The delinquency specialization hypothesis predicts that a factor analysis of delinquency data should yield clear-cut factors representing patterns of delinquent behavior, while the generalized delinquency hypothesis predicts that the first factor extracted should be by far the largest; this factor should be a general delinquency factor. Both the specialized and generalized delinquency theories predict that systematic patterns of behavior can be found in delinquency data. In contrast, the random delinquency hypothesis predicts that factor analysis should find mostly unique variance. Although no one theory is consistently supported in the literature, the currently available evidence indicates some kind of systematic patterning in delinquency, whether specialized or generalized. Lack of consensus about the nature of these patterns is attributed largely to instrument differences. A factor analysis of official arrest records for nearly 29,000 male members of a Danish birth cohort included 56 common or criminologically interesting offenses. Four factors were found: (1) general crime, (2) traffic-related offenses, (3) white-collar crime, and (4) sex offenses. Although the analysis resulted in a large unique variance component, the four factors cross-validated quite well. The issue of difficulty factors as a competing explanation was addressed, and it is concluded that the present results are not the product of a difficulty artifact. Results support the generalized delinquency hypothesis; however, many crimes appear to be independent of any pattern. Tabular data and 20 references are provided. (Author abstract modified)