Objective: Numerous operationalizations of persistent offending have been used in the extant research with the assumption that these findings are generalizable. We tested this assumption by comparing the criminal careers of persistent offenders identified by different methods. Method: We examined 38 operationalizations of persistent offending and the groups they identified. Criminal careers were measured using official conviction data from the Cambridge Study in Delinquent Development (n = 411). Results: The groups of persistent offenders differed in prevalence rates (ranging from 1.24% to 29.53% of the sample), average age of onset (from 10.86 to 26.07 years), average criminal career duration (from 16.96 to 39.86 years), average convictions (from 4.03 to 23.33, λ ranged from 0.3 to 1.17 convictions per year), and offender overlap (from 0% to 100%, ORs ranged from 0.34 to 787.5). Persistent offenders identified by the Group Based Trajectory Model were least likely to be identified by any other operationalization. Conclusion: Different operationalizations generally identified qualitatively different offenders as persistent, suggesting that the findings on persistent offending in the literature may not be generalizable across studies. However, our analyses are limited to the CSDD, and so further research is needed.