Background The key question is: are self‐reports and official records equally valid indicators of criminal offending? Aims We examine the correspondence between self‐reports and official records of offending, the similarity of childhood and adolescent individual and contextual predictors of both measures of offending, and the similarity of age 48 correlates of both measures of offending. Methods Men (N = 436) from the Columbia County Longitudinal Study, a sample of all 3rd graders in Columbia County, New York, in 1959–60, participated. The youth, their peers and their parents were interviewed when the youth were age 8; the youth were later interviewed at ages 19, 30 and 48. Results We found moderate to high correspondence between self‐reports of having been in trouble with the law and official arrest records. Lifetime self‐reports and official records of offending were generally predicted by the same childhood and adolescent variables, and were correlated with many of the same adult outcome measures. By age 48, life‐course non‐offenders defined by either self‐reports or official records had better outcomes than offenders. Conclusions The results validate the use of adolescent and adult self‐reports of offending, and the early identification of individuals at risk for adult criminal behaviour through childhood parent and peer reports and adolescent self and peer reports. Copyright © 2014 John Wiley & Sons, Ltd.