This paper reports on the results of a psychological study conducted in Ontario, Canada, that attempted to answer the question of why some people develop gambling problems while others do not. A group of social gamblers (n = 38), sub-clinical problem gamblers (n = 33) and pathological gamblers (n = 34) completed a battery of questionnaires. Compared to non-problem gamblers, pathological gamblers were more likely to report experiencing big wins early in their gambling career, stressful life events, impulsivity, depression, using escape to cope with stress and a poorer understanding of random events. We grouped these variables into three risk factors: cognitive/experiential, emotional and impulsive and tested the extent to which each risk factor could differentiate non-problem and pathological gamblers. Each risk factor correctly identified about three-quarters of the pathological gamblers. More than half (53%) of the pathological gamblers had elevated scores on all three risk factors. Interestingly, 60% of the sub-clinical cases had elevated scores on only one risk factor. The results are interpreted in terms of a bio-psycho-social model of gambling addiction.