The dissertation attempts to test empirically social control, differential association and self-concept theories of delinquency, using data from a national probability sample. This involves a four-step process. First, an examination of social control theory and differential association contrasts the theories on important issues. Second, the resulting hypotheses form a structural equation model. Third, measurement error in important explanatory contracts is modeled and assessed, and statistically controlled using confirmatory factor analysis. Finally, unobservable latent variables corrected for attenuation due to unreliability tie into a causal model of social control and delinquent behavior. Parameters of this model come from estimates generated by maximum likelihood procedures. Data from the project contain indicators of social control theory that have reasonable measurement properties. Confirmatory factor analysis control the measurement error of these indicators. Two of the three hypotheses derived from social control theory do not prove valid. Differential association theory receives only modest support. Both differential identification and self-concept theories of delinquency produce negative evidence. Both theories fail to specify the nature and length of the causal lag between explanatory concepts. This makes empirical testing using longitudinal data difficult. A situational explanation of the delinquent act might overcome this problem. Notes, tables, and figures accompany the text. Seventeen pages of references are included. (Author summary modified)