In the extant literature, adult-onset offending has usually been identified using official sources. It is possible, however, that many of the individuals identified would have had unofficial histories of prior offending. To investigate this issue, the men from the Cambridge Study in Delinquent Development (CSDD) were examined. The CSDD is a prospective longitudinal study of men from inner-city London, followed from age 8 to age 48. Onset of offending was identified using official records and then the self-reported offending of the adult-onset offender group (with a first conviction at age 21 or later) was compared to others. All the adult-onset offenders self-reported some previous offending in childhood and adolescence but most of this offending was not sufficiently frequent or serious to lead to a conviction in practice. About one-third of adult-onset offenders were considered to be self-reported delinquents who were realistically in danger of being convicted because of the frequency of their offending. For some, the adjudication by the criminal justice system was simply the first time that their ongoing pattern of offending had been detected. Their lack of detection was because the types of offences they were committing had lower detection rates.