EDMONDS, J.
Father and mother appeal a judgment terminating their parental rights in their then 21-month-old son Matthew. They argue that the state failed to prove by clear and convincing evidence that they are unfit parents, that reintegration into their home is improbable within a reasonable time because they are unlikely to change, and that termination of their parental rights was in Matthew’s best interest. ORS 419B.504; State ex rel SOSCF v. Stillman, 333 Or 135, 145, 36 P3d 490 (2001). On de novo review, ORS 419A.200, we reverse.
This case involving Matthew is a sequel to State ex rel Juv. Dept. v. Nguyen, 182 Or App 294, 48 P3d 864 (2002), aff'd in part, vac’d in part, 335 Or 255, 99 P3d 1025 (2003) (Nguyen 7). In that case, we reversed the trial court’s ruling denying the state’s petition to terminate father’s and mother’s parental rights to Matthew’s older sisters, Mary and Martha. Our decision was predicated on the finding that one of the parents had inflicted serious abuse on Martha, repeatedly breaking her limbs and fracturing her skull; that the other parent was aware of that conduct; and that neither party acknowledged responsibility. The Supreme Court allowed review of our decision, but before an opinion was issued on the merits, father, mother, and the state reached a stipulation under which the parents voluntarily terminated their parental rights as to Mary and Martha. State ex rel Juv. Dept. v. Nguyen, 335 Or 255, 66 P3d 1025 (2003).
In this case, the state argues:
“Indeed, if anything, the mother and father now are less fit to care for a young child than they were at the time of the proceedings to terminate their rights to Martha and Mary. That is so because, since those proceedings, this court — on de novo review of the judgment denying termination in Martha’s and Mary’s case — found by clear and convincing evidence that either the mother or father severely and repeatedly abused Martha, and yet, despite that finding (which is a matter of public record), the mother and father remain together, they continue to maintain that they do not know how or why Martha was abused, and they have refused to engage in the process of identifying and curing [608]*608the problems that led to Martha’s injury. Accordingly, the termination judgment should be affirmed.”
(Some emphasis in original; citations omitted.)
On the other hand, father and mother argue,
“Since the beginning of the case, the state has demanded that the parents provide an explanation for the injuries sustained by Martha (i.e., a parent must admit to causing the injuries or acknowledg[e] a failure to protect the child from the abusing parent). The state has not accepted or found credible any other explanation or account of what may have happened to the child offered by the parents. The state contends that absent an admission by [parents] or acknowledgment of the harm or for the failure to prevent the harm, the parents cannot present a safety plan sufficient to insure the safe return of Matthew to the parents’ home.
“The parents cooperated with the state and the investigation of the circumstances of the injuries sustained by Martha. Over a period of more than three years, the parents engaged in numerous services, such as parenting classes, visitation, psychiatric and psychological evaluations and individual and couples’ counseling. The parents have consistently demonstrated a strong commitment to safely care for their children. At the time of this termination trial, the evidence was unequivocal that the parents exceeded the expectations of the agency in terms of compliance with services. The evidence also demonstrated the parents^] ability to provide minimally adequate care to Matthew for the period of time from birth to his removal (a period of more than ten months). The parents further demonstrated a capability of instituting changes in their conduct and circumstances in order to insure the proper care for and safety of the child when returned to their home.
“The [trial] court failed to adequately assess the parents’ ongoing participation in services that has ameliorated the conditions from which the child was removed. The behavior of both parents changed over the intervening period of time between the prior determination [sic] trial and the subsequent termination proceeding as well as a result of their ongoing work in therapy, parenting and other social services.”
[609]*609Faced with those competing arguments, we turn to the facts and the disposition of Nguyen I, which form the background to the present case. In September and October 1999, after doctors discovered suspicious injuries on four-month-old Martha Nguyen during a routine medical exam, state authorities began an investigation to determine whether Martha had suffered abuse. They also began meeting with father and mother and providing services such as parenting classes. The state, meanwhile, petitioned the juvenile court to take jurisdiction of Martha and her older sister Mary. The court found the children to be within its jurisdiction and ordered the parents to undergo therapy and participate in other forms of treatment. It also denied their request to retain physical custody of the children.
Father and mother followed the service agencies’ recommendations, but they did not provide any plausible explanation for how Martha received her injuries. In September 2000, the state filed a petition to terminate their parental rights to the children. Trial on the petition occurred in July 2001; the trial court dismissed the petition, finding that
“there is no reason to believe that [mother and father], or either of them, is incapable of parenting. The Court is fully cognizant of the serious nature of the injuries that the younger child sustained. While child abuse continues to appear to be the most likely explanation of the injuries, the fact remains that these are, by all accounts, extraordinarily capable and fit parents who are open to change, to training and are fully competent to parent.”
(Brackets in original.) Almost immediately thereafter, on August 11, 2001, Matthew was born. Two days later, the state sought to remove him from parents’ home; the court denied the request but imposed a comprehensive safety plan requiring extensive in-home supervision and additional parenting skills training. In October 2001, the court took jurisdiction over Matthew, but he remained in the physical custody of his parents.
The state, meanwhile, appealed the trial court’s dismissal of the termination petition involving Martha and [610]*610Mary. We decided that case on June 26, 2002. On de novo review, we concluded that,
“[o]n this record, the evidence is clear and convincing that one parent caused Martha’s injuries and that the other was either unaware of or did nothing to prevent it.
“Both parents argue that, even if one of them were unfit in the past, the state still failed to prove that they are presently unfit. They argue that they ‘have embraced mental health therapy and parenting resources, as if one of them [had] caused the injuries.’ We acknowledge that mother and father have participated in and successfully completed numerous services. For example, mother and father successfully completed a 12-week course that addressed different parenting topics for one hour each week. While the parents’ participation in that course, as well as others, has been exemplary, completing generalized course work does not address the specific problem that caused Martha’s injuries. If, as we find, either mother or father repeatedly broke their four-month-old daughter’s bones, the expert testimony in this case establishes that mother and father must acknowledge the problem, identify its cause, and work to correct it. Unless and until the parents take those steps, and they have not done so, their generalized participation in services does nothing to cure the specific problem that led to Martha’s injuries. We find that mother and father are not presently fit.”
Nguyen I, 182 Or App at 305-06 (brackets in original; internal citations omitted). We also found that the conduct rendering them unfit was unlikely to change and that termination of their parental rights was in the children’s best interest. Id. at 306-07. We reversed and remanded with instructions to enter a judgment terminating their parental rights. Id. at 307.
Matthew was removed from his parents’ home the following day. At that time, he had been living with his parents under the court-approved safety plan for approximately 10 months. Two days later, the court denied parents’ request to return Matthew to their physical custody. On August 23, 2002, the state filed a petition to terminate father’s and mother’s parental rights to Matthew.
[611]*611While the resolution of Matthew’s termination case was pending, the parents petitioned for Supreme Court review of Nguyen I. The case went into the appellate settlement program, and the parties reached a stipulated agreement that resulted in the voluntary termination of both parents’ rights in Mary and Martha. On March 12, 2003, the Supreme Court allowed parents’ petition for review, affirmed Nguyen I except for that aspect of the decision concerning instructions to the trial court on remand, vacated those instructions, and remanded with instructions “to enter a judgment terminating both parents’ rights and indicating that the termination of their parental rights is voluntary.” Nguyen 1, 335 Or at 255.
Matthew’s termination hearing began on March 17, 2003. The evidence indicated that mother and father had used all of the services mandated by the safety plan and that mother had attended additional parenting skills training at her own initiative. Yen Le, who had lived in the home with the family for almost two years in order to observe the parents’ behavior with the children, testified that she never saw mother harm Matthew and that parents appeared to love each other and love the children. The testimony from other service providers was similarly positive. Also at trial, the state offered the record of the trial in Nguyen I, the transcript of the protective hearing held in August 2002, and the testimony of a number of witnesses, including two expert witnesses, Eric Johnson and Renee Daniel Hershey. The following is a summary of their testimony.
Johnson holds a doctoral degree in psychology. Since 1993, he has been in private practice, a practice devoted exclusively to forensic matters in the areas of child abuse and neglect. He has never interviewed parents or treated them professionally.1 He conceded that there are no scientific studies that address the issue of the risk of repeat abuse where there is an absence of acknowledgment of abuse. Rather, all [612]*612of the available studies address persons who are acknowledged offenders. He testified at trial, as he had in the August hearing, that he was unable to perform a risk assessment regarding the likelihood of parents abusing Matthew, based on the information available to him, principally because of their unwillingness to acknowledge the abuse of Martha. In the absence of such an acknowledgment, the risk assessment would be high, in his view, until proven otherwise.2 He also said that, based on the psychological evaluations of parents that he had reviewed, “they were not shown to have significant mental health-related problems.”
In the August hearing, when asked whether the 10-month period of time during which parents had custody of Matthew had any bearing on his assessment of risk, he replied:
“The fact that there has been an extended period of time, and I consider 10 months to be a noteworthy period of time, no apparent additional problems of this type suggest that any original problems are unlikely to have been caused by a severe or chronic underlying problem. It does not rule out situational problems, it does not rule out problems that could be caused by acute stress, up to and including depression and postpartum depression. It does not rule out those kind[s] of causes. So it is helpful in narrowing the field, but [613]*613it does not rule out, you know, the problem that this could reoccur.”
When asked whether “there is always going to be a level of risk” until the time that “there is an admission or an accountability taken for an act,” Johnson responded:
“What I would say is until you have all the information, you cannot conduct an adequate risk assessment; and until you have an adequate risk assessment, you cannot put forward an adequate treatment plan; and until you have an a treatment plan and parties that are agreed upon what the treatment needs to focus on, you do not have treatment. And under these conditions, you cannot guarantee the safety of any child in a parent’s care under those conditions.”
(Emphasis added.)
Significantly, in its brief to the court in this case, the state does not expressly point to any testimony by Johnson at the August 2002 hearing or at the trial that assesses the risk to Matthew presented by parents at the time of trial. Rather, it relies on the following statement made by this court in Nguyen I:
“Dr. Eric Johnson, a psychologist with extensive experience treating child abusers, explained that a person must acknowledge having abused a child before the person can be treated. Unless and until the problem that caused the abuse is identified, it cannot be remedied. As Johnson explained, ‘If you don’t know what the problem is, you cannot enter into treatment. You cannot undertake effective treatment. You don’t know what your are treating.’ Accordingly, he concluded that it is ‘vitally important’ for the abuser to acknowledge that he or she abused the child and to begin working on the problem that caused the abuse.”
Nguyen I, 182 Or App at 303-04 (footnote omitted).
Daniel-Hershey testified both at the August 2002 hearing and at trial. She is a licensed clinical social worker with a master’s degree in social work and the codirector of a mental health agency that serves children in foster care awaiting adoption. She has never interviewed parents or treated them. She was unaware of any scientific studies that correlate the risk of future abuse with the failure to acknowledge existing abuse. She was asked whether the risk of harm [614]*614to Matthew had decreased because of the ten months that he was in parents’ custody. She replied:
“I don’t believe that we have enough information to assess risk. And I really think that how you would mitigate the risk for a child is that you would understand what happened, and then, again, you look at the event, you look at what triggered the event, and then you do your best to put in certain protective factors that would prevent that from happening again.”
In her view, parents’ “track record” of caring successfully for Mary and Matthew was “irrelevant” because it “did not protect Martha.” As a professional therapist, she believes that a person cannot enter into meaningful therapy until the person says, “I did it[,]” although she acknowledged that other therapists might hold different professional views. She recommended putting Matthew in foster care “until there’s a definite decision about what his permanent plan is” because of a concern regarding an attachment disorder caused by multiple changes in custody.
In its brief, the state also relies on testimony from expert witnesses called by parents. Those witnesses include Lorah Sebastian. She holds a doctorate in educational psychology. Her present practice consists of working with children and families in the evaluation and treatment of a variety of situations including the physical abuse of children. She has also testified on behalf of the State of Oregon in a number of dependency and termination hearings. She, too, has never met with the parents or treated them professionally. She, too, was provided with parts of the transcript of the first termination trial and psychological evaluations of parents. The state says in its brief, “Similarly, Dr. Lorah Sebastian testified that a therapist must know what he/she is treating to target the problem.” (Internal quotation marks omitted.) Significantly, the following colloquy occurred at trial when Sebastian was on the witness stand:
“Q. * * * And based on your review of those documents, are you able to in this case to tell us what the risk would be to the child Matthew if he were to return to his parents?
“A. No.
[615]*615“Q. And why not[,] even though you’ve reviewed all those documents and you’re an expert in these matters, why can’t you tell us what the risk is?
“A. Well, a number of reasons why I couldn’t tell you. But probably the first one is that I’ve never seen Matthew, I’ve never seen the parties, the parents in this case with Matthew or interview them or evaluated them[.]”
Later, Sebastian was asked by counsel,
“Q. In a given case[,] if I were to give you three factors of a case of child abuse and ask you to assign a level of risk about the recurrence of child abuse, if I were to identify the fact that the parents were Vietnamese, the fact that there had been no acknowledgment of the abuse, and the fact that there was significant, serious or significant injuries to the child, would those factors, isolated from any other factors, allow to you to assign a risk level in that particular case?
“A. Without seeing the parties?
“Q. Without seeing the parties.
“A. No, especially not, because of the uniqueness. Those are fairly unique and reasonably unusual set of co-occurring circumstances that would — I would think make it even more difficult and would have more need for a more thorough look at things, as thorough as you could be, to determine an answer to the question.”
The state also relies on the testimony of Paul Leung, another expert witness called by parents. According to the state, Leung testified that “without further intervention^] there is risk of further abuse.” (Internal quotation marks omitted.) Leung, an associate professor of psychiatry at the school of medicine at Oregon Health Sciences University (OHSU), is a medical doctor who has been active in the IndoChinese psychiatric program at OHSU since 1981. He has worked directly with families of Vietnamese heritage on a weekly basis since that time. He also speaks Vietnamese fluently. Mother is one of his patients, and he had been treating her for approximately three-and-one-half years at the time of trial. He estimated that he had seen her approximately 20 times in that time period. He found no evidence that mother suffers from a psychosis or a personality disorder. He also said that he is unable to identify any condition that could [616]*616have contributed to violent behavior toward Martha. He testified that assessing the risk of future violent behavior is “part of our routine * * * practice.”
Leung’s testimony at the first trial is described in Nguyen I as follows:
“Mother was evaluated by Dr. Paul Leung, who was also chosen in part for his familiarity with Vietnamese culture. Leung concluded that mother ‘enjoyed a very healthy mental [and] emotional condition’ before the discovery of Martha’s injuries and removal of her children and that she was capable of providing adequate parenting for her daughters. Leung explained, however, that, if mother had caused Martha’s injuries ‘out of stress’ and being overwrought,’ he would ‘have very much concern about the ability of this woman to care for these two children’ and would recommend ‘additional treatment to the mother [and] the father.’ ”
182 Or App at 300 (brackets in original; footnote omitted).
At the trial in this case, Leung testified that
“it would be less of a risk for Matthew to return to the parents, because my contact with Matthew in the clinic, I saw a thriving, chubby, happy baby and that — I took that into the consideration^] as far as Matthew staying with parents[.]”
Later, Leung was asked whether, based upon his clinical experience, he believed that it is necessary for a person to acknowledge misconduct such as child abuse in order to accept and participate in effective treatment. He replied,
“My whole clinical experience, no, it is not necessary to have to be that way. Many * * * patients that I personally have treated in sexual abuse cases, in alcohol and drug addiction cases, likely at the beginning of treatment the patient would have a certain degree of resistance to even accept what really is a fact in front of the person.
“But as treatment proceeded], the trusting level increases between the therapist and the patient. The chance is that the patient would come to a point to accept what had happened and admit to it. That would certainly bring forth meaningful therapy. But I could not find any support in my own reading or research and talking to my colleagues that unless a person openly publicly admit to [617]*617something has been done wrong, then you cannot meaningfully treat the person. I just don’t — I can’t get that. So that’s the best I can put it.”
(Emphasis added.) On cross-examination, Leung conceded that his diagnosis depends in great part on the information provided by the patient and that, in formulating risk assessments, it is important to understand the underlying causes of abuse
“so that a plan can be formulated to help the person to look at the situation and at least learn how to deal with * * * that in the future when similar conditions arise, at least the person would have the skills to deal with it instead of just resort to automatic reaction and abuse would come up again.”
The parents also offered the testimony of their other treatment providers including Cach Nguyen, a member of the Multicultural Counseling Professional Consortium. He has known the parents since March 2000 and continued to counsel father for a period of two-and-one-half years until December 2002, when the funding for his services was eliminated. Nguyen testified that he had not discovered any anger management, psychological, or psychiatric issues on the part of father while counseling him. While Matthew was in the family home, Nguyen observed father with him seven or eight times. He believed that father and Matthew were bonded and that mother was a caring mother who loved Matthew very much.
The parents also called as a witness Hoang Ta, a community health nurse employed by Multnomah County. She speaks Vietnamese and has had training to observe signs of child abuse. Ta visited parents’ home approximately 30 to 50 times. She began working with parents when mother was about three or four months pregnant with Matthew and continued until he was removed from the family home. From her observations, she concluded that parents took good care of Matthew, that there were no signs of abuse, and that she saw nothing that caused her concern about his safety. In the August hearing, she recommended that Matthew be returned to his parents because she had not detected any [618]*618abuse, his growth and development were normal, and there was significant bonding between him and parents.
Kim Pham also testified. She has been a counselor and a research assistant with the OHSU Intercultural Psychiatric Program for ten years. She has a masters degree in counseling and is fluent in Vietnamese. She has worked with mother for over two years, administering individual therapy for depression attributed by Pham to the termination of mother’s parental relationship with Mary and Martha. According to Pham, mother has made progress, insofar as her depression is concerned. Mother continued in therapy after Matthew was born, bringing him to treatment sessions, and Pham did not observe anything about Matthew that caused her to be concerned about the care that he was receiving from mother.3
ORS 419B.500 and ORS 419B.504 set out a multipart standard that the state must meet before it can terminate parental rights. Stillman, 333 Or at 144-45. The state must prove by clear and convincing evidence that, at the time of the termination trial, the parents are presently “unfit by reason of conduct or condition seriously detrimental to the child” and integration into the parents’ home is “improbable within a reasonable time” because the conduct or condition is “not likely to change.” Id. at 145-46 (internal quotation marks omitted). If the parents are presently unfit and integration is not probable within a reasonable time, the remaining question is whether termination is in the child’s best interests. ORS 419B.500.
We turn first to the state’s argument on appeal, as well as the gist of the trial court’s decision, that, because parents have not acknowledged their responsibility for the abuse of Martha, nothing has changed since the termination of parents’ rights in Mary and Martha. The state relies on the statement that we made inNguyen I, quoted previously, that, until the parents acknowledge their responsibility for [619]*619Martha’s injuries, they are unfit to parent Matthew. Consequently, the state concludes that our statement in Nguyen I ought to control the result in this case because father and mother continue to not accept responsibility for Martha’s injuries.
What is troubling about the state’s reasoning is that the statute that governs this case requires the state to demonstrate by clear and convincing evidence that father and mother are presently unfit to be the parents of Matthew. The state’s reasoning implies that it is incumbent on father and mother to demonstrate a change from the circumstances that existed in Nguyen I. See ORS 419B.521. Such reasoning incorrectly places the burden of proof in this case on father and mother. The short answer to the state’s argument is that ORS 419B.504 requires the state to prove by clear and convincing evidence that father and mother were presently unfit in March 2003, not that they were unfit in July 2Ó01, the time of our decision in Nguyen I. While the record in Nguyen I is relevant to our assessment in this case, and while the parents continued in their failure to acknowledge responsibility as of March 2003, a fact also relevant to our decision, we are required under the applicable statutes to consider also the evidence about what has occurred since July 2001. In other words, we are confronted with a different evidentiary record in this case than existed in Nguyen I, and we must again apply the statutory standard regarding the state’s burden of proof to what is, in effect, a combined evidentiary record of what occurred in Nguyen I and what has occurred since.
This case is not an easy case to decide. There is a legitimate concern regarding the safety of Matthew in light of the unexplained injuries to Martha. The difficulty in deciding this case is exacerbated by the inherently speculative process of predicting a risk of future harm, a process that the state’s three expert witnesses expressly refused to undertake regarding Matthew’s situation. Ultimately, however, we must ask whether the state carried its burden to prove by clear and convincing evidence that father and mother were presently unfit to be Matthew’s parents as of the date of trial in 2003. The “clear and convincing” evidence standard requires that the state must show “that it is highly probable that father and mother are not presently able, or will not be [620]*620able within a reasonable time, to meet the physical and emotional needs” of Matthew. State ex rel Juv. Dept. v. Johnson, 165 Or App 147, 156, 997 P2d 231 (2000) (emphasis in original).4
The cornerstone of the state’s argument in this case is that father and mother have not acknowledged responsibility for the abuse of Martha. However, the state’s expert witnesses and the parents’ expert witnesses disagree on the significance of the need for such an acknowledgment. The state’s witnesses are uniform in their belief that no effective rehabilitative treatment can occur without such an acknowledgment. It follows, in their view, that the three-and-one-half years of continued therapy undertaken by father and mother is of little relevance. Also irrelevant, in their view, is the fact that father and mother cared for Matthew in an exemplary fashion for 10 months during the same time period in Martha’s life during which she was injured. It follows, then, according to those experts, that the risk of injury that existed in July 2001 continues unabated as of May 2003. The flaw in the state’s argument is that it seeks to carry its burden to prove that it is highly probable that father and mother present a risk of harm to Matthew based on its experts’ concessions that they did not have enough information to assess the risk of harm at the time of trial.
In contrast, the expert witnesses who have treated father and mother and interacted with them throughout the court proceedings in this case and the previous cases believe, contrary to the state’s allegations, that conditions giving rise to the previous termination proceedings have ameliorated since that time and that the risk to Matthew has lessened. Both the state’s and parents’ expert witnesses agree that [621]*621father and mother do not suffer from any psychosis or personality disorder that could have prompted abusive conduct towards Martha on their part. They all agree that the abuse of Martha must have occurred under situational circumstances. Finally, no expert witness in this case has made an affirmative risk assessment that posits that parents will cause harm to Matthew in the future. On the other hand, no expert witness, including those who support father’s and mother’s position, is able credibly to guarantee that there is no risk of harm to Matthew in the future.
So, how to weigh the evidence in this case? The fact of the potential of Matthew being harmed if he is returned continues to loom in this case, based on what happened to Martha. On the other hand, we are required to determine whether that potential is highly probable. The state’s experts’ testimony that they are unable to make a risk assessment and so arrive at a conclusion based upon the most conservative approach tailored to Matthew’s safety is not to be faulted; but also, they are not assigned the task of deciding this case based on the requirements of the statutes and the burden of proof assigned to the state. As for the expert witness who have treated father and mother, although they concede that a risk of harm remains, they recommend the return of Matthew to parents; thereby, belying any notion that the risk of harm is highly probable. What about the fact that parents are unwilling to acknowledge responsibility for Martha’s injuries? How much weight should that fact be given? The state’s experts declare that fact to be the lynchpin to father’s and mother’s successful treatment. But those experts have not treated father and mother. On the other hand, Dr. Leung, a highly qualified psychiatrist, who works directly with families of Vietnamese heritage, speaks Vietnamese fluently, and has treated mother for approximately three-and-one-half years, testified that an acknowledgment of responsibility is not critical to effective therapy for mother. In addition, the community health nurse who monitored Matthew’s care on over 30 occasions recommended to the trial court that Matthew be returned to father and mother.
In sum, we are left with a record in conflict where the standard of proof requires that the conditions of parents alleged by the state be proved by evidence that demonstrates [622]*622that the risk of harm to Matthew is highly probable if he is returned to them. In light of the testimony of nurse Ta, therapist Nguyen, and psychiatrist Leung — all professionals in their fields of expertise; all people who have treated or have had contact with father and mother; all people who are familiar with father’s and mother’s culture; and all of whom essentially support the proposition that father’s and mother’s parental rights should not be terminated — we conclude that the state has not met its burden of persuasion. We therefore reverse the decision of the trial court.6
Reversed and remanded.