Path analytic procedures are employed to test the most explicit causal paradigm of the self-concept-delinquency relation: the delinquency-as-self-enhancement model. Despite: (1) variations in the types of delinquency analyzed, (2) restrictions placed on the sample in testing for possible contingent relations, and (3) variations in the time lags used to compute the regression coefficients, no substantial effect of self-esteem on subsequent delinquency was found when the effects of prior causal variables were partialled out. Additionally, no evidence of a self-enhancement effect (i.e., a positive effect of delinquency on subsequent self-concept) was discovered. These results are discussed in relation to prior research and social policy assumptions which focus on self-images as a key variable for reducing delinquency. (Author abstract)