Literature on psychological characteristics of abusive families suggests that psychometric assessments can be developed to screen parents for abusiveness. Advocates of primary prevention of child abuse have encouraged the development of such perinatal screening devices. However, social values define abusiveness, making its assessment especially prone to error. If a nearly perfect test existed with a sensitivity of .99 and a specificity of .99, half of the parents identified as abusive would not be abusive. Existing tests have done far worse, achieving at best a sensitivity of .77 and a specificity of .83. Although the low base rate of abuse prevents standard validity coefficients from accurately portraying the validity of tests for abusiveness, current tests appear to have little validity and high rates of predictive error. Given various levels of test precision, probabilities for correct and incorrect assessments of abusiveness can be determined; errors in virtually any test for abusiveness will be false positives, i.e., a detection of abusiveness when none exists. Considering the ethical and economical issues involved in testing the general population and intervening upon nonabusive families identified as abusive as well as the obstacles to gaining predictive validity in tests, an abandonment of testing efforts warrants serious consideration. (Author/NRB)