OBJECTIVE: To identify patterns of sports gambling that discriminate sports bettors with self-reported gambling-related problems from sports bettors without such difficulties. METHODS: Secondary data analysis of the actual betting behavior observed during the first 2 years of a longitudinal study of 47,134 subscribers to an Internet sports gambling site. This sample included the gambling behavior of 679 bettors who self-reported the reason for closing their accounts during that period. We contrasted the behavior of those who closed their accounts because of gambling-related problems (n = 215, 32%) to the behavior of other account closers (ACs) who were either not satisfied with the service (n = 113, 17%) or no longer interested in betting (n = 351, 52%). RESULTS: Exploratory multivariate discriminant function analyses identified a sub-group of approximately half the ACs with gambling-related problems who exhibited a homogeneous and distinct pattern of sports-betting behavior. Compared to other ACs, this sub-group made more and larger bets, bet more frequently, and were more likely to exhibit intense betting soon after enrollment. The group estimation formula derived from this prototype applied to an independent sample of ACs confirmed the prevalence of this distinct gambling pattern. CONCLUSION: Because Internet gambling provides a unique opportunity to study actual gambling behavior, it is possible to identify betting patterns that can lead to the development of gambling-related problems. This pattern recognition can inform the development of interventions to help disordered gamblers recognize their risky behavior and avoid further problems.