A survey of research findings notes that high individual crime rates are predicted by an early age at the onset of offending, a serious first offense, and a high past crime rate. Other predictors are early antisocial behavior, convicted parents and siblings, low family income, school failure, poor employment record, and drug use. The paper contends that prediction of individual crime rate should be investigated in prospective longitudinal surveys and that predictor and criterion variables be selected on theoretical grounds rather than availability in records. Also needed is a standard index of predictive efficiency. The conclusion explores ethical, legal, and practical issues raised by the use of prediction information in criminal justice decisions. Over 100 references. (Author abstract modified)