While population surveys have been carried out in numerous jurisdictions internationally, little has been done to assess the relative strength of different risk factors that may contribute to the development of problem gambling. This is an important preparatory step for future research on the etiology of problem gambling. Using data from the 2006 California Problem Gambling Prevalence Survey, a telephone survey of adult California residents that used the NODS to assess respondents for gambling problems, binary logistic regression analysis was used to identify demographic characteristics, health-related behaviors, and gambling participation variables that statistically predicted the odds of being a problem or pathological gambler. In a separate approach, linear regression analysis was used to assess the impact of changes in these variables on the severity of the disorder. In both of the final models, the greatest statistical predictor of problem gambling status was past year Internet gambling. Furthermore, the unique finding of a significant interaction between physical or mental disability, Internet gambling, and problem gambling highlights the importance of exploring the interactions between different forms of gambling, the experience of mental and physical health issues, and the development of problem gambling using a longitudinal lens.