This study explored several approaches to the prediction of probation violation or nonviolation in 287 male juvenile delinquents who had never been institutionalized. Predictions from a base expectancy table were significantly though not greatly better than chance. Clinical judgments from MMPI profiles by 2 independent judges completely failed to improve upon chance, as did actuarial predictions based on the delinquency-proneness data of Hathaway and Monachesi. An attempt to employ MMPI responses in the construction of a moderator scale to enhance prediction from the base expectancy data also failed. The failure of prediction from the MMPI data is consistent with Sutherland's differential association theory of crime causation.