Research indicates that specific types of gambling, such as electronic gaming machines (e.g., Video Lottery Terminals (VLTs), slot machines, virtual gaming machines), are associated with increased odds of experiencing gambling problems. Recent findings suggest that to advance our understanding of gambling-related problems scientists need to focus less on a simple association (e.g., specific gambling type) and more on complex models that include the extent of gambling involvement. The objective of this study is to advance this area of investigation by establishing the generalizability of the involvement effect to the general population of Canada, as well as to examine two potential moderating factors: gender and age. Secondary data analysis of the nationally representative Canadian Community Health Survey (CCHS) cycle 1.2 (data collected during 2002, response rate = 77 %, n = 18,913) using logistic regression models were conducted. All types of gambling were associated with problem gambling. However, when adjusting for gambling involvement (i.e., the number of games played during the past year), these specific game relationships were either eliminated or attenuated. Significant relationships remained for instant win lottery tickets, bingo, card and/or board games, electronic gaming machines outside of casinos, electronic gaming machines inside casinos, other casino gambling, horse racing, sports lotteries, and games of skill. For many types of gambling, the nature of the relationships seemed to be a function of the frequency of engagement with specific games. Gender and age did not moderate these findings. These findings indicate that focusing on a narrow direct cause (e.g., game type) for gambling problems needs to shift towards a more complex model that also includes the level of gambling involvement.