Most parents who live apart negotiate custody arrangements on their own or with the help of lawyers, mediators, or other professionals. However, psychologists and other mental health professionals increasingly have become involved in evaluating children and families in custody disputes, because of the large number of separated, divorced, and never-married parents and the substantial conflict that often accompanies the breakup of a family. Theoretically, the law guides and controls child custody evaluations, but the prevailing custody standard (the "best interests of the child" test) is a vague rule that directs judges to make decisions unique to individual cases according to what will be in children's future (and undefined) best interests. Furthermore, state statutes typically offer only vague guidelines as to how judges (and evaluators) are to assess parents and the merits of their cases, and how they should ultimately decide what custody arrangements will be in a child's best interests. In this vacuum, custody evaluators typically administer to parents and children an array of tests and assess them through less formal means including interviews and observation. Sadly, we find that (a) tests specifically developed to assess questions relevant to custody are completely inadequate on scientific grounds; (b) the claims of some anointed experts about their favorite constructs (e.g., "parent alienation syndrome ") are equally hollow when subjected to scientific scrutiny; (c) evaluators should question the use even of well-established psychological measures (e.g., measures of intelligence, personality, psychopathology, and academic achievement) because of their often limited relevance to the questions before the court; and (d) little empirical data exist regarding other important and controversial issues (e.g., whether evaluators should solicit children's wishes about custody; whether infants and toddlers are harmed or helped by overnight visits), suggesting a need, for further scientific investigation. We see the system for resolving custody disputes as deeply flawed, for reasons that go beyond the problem of limited science. The coupling of the vague "best interests of the child" test with the American adversary system of justice puts judges in the position of trying to perform an impossible task, and it exacerbates parental conflict and problems in parenting and coparenting, which psychological science clearly shows to be key factors predicting children's psychological difficulties in response to their parents' separation and divorce. Our analysis of the flawed system, together with our desire to sharply limit custody disputes and custody evaluations, leads us to propose three reforms. First, we urge continued efforts to encourage parents to reach custody agreements on their own-in divorce mediation, through collaborative law, in good faith attorney negotiations, in therapy, and in other forums. Some such efforts have been demonstrated to improve parent parent and parent-child relationships long after divorce, and they embrace the philosophical position that, in the absence of abuse or neglect, parents themselves should determine their children's best interests after separation, just as they do in marriage. Second, we urge state legislatures to move toward adopting more clear and determinative custody rules, a step that would greatly clarify the terms of the marriage contract, limit the need for custody evaluations, and sharply narrow the scope of the evaluation process. We find particular merit in the proposed "approximation rule" (recently embraced by the American Law Institute), in which postdivorce parenting arrangements would approximate parenting involvement in, marriage. Third anal finally, we recommend that custody evaluators follow the law and only offer opinions for which there is an adequate scientific basis. Related to this, we urge professional bodies to enact more specific standards of practice on this and related issues.