Although it is often assumed that electronic gaming machines (EGMs) are associated with the highest level of risk, it has proved difficult to find reliable evidence in support of this proposition. In this paper, we analysed statistics from major Australian community prevalence studies for the period 2011–2020 to investigate whether EGMs (in comparison to racing and casino table games) have a stronger association with problem gambling. All prevalence studies reviewed used telephone sampling and the Problem Gambling Severity Index to assess problem gambling. In this paper, we examine the principal hypothesis using several lines of evidence, including whether problem gamblers are more likely to gamble and gamble regularly on EGMs as opposed to racing and casino games and if the EGM-problem gambling association was maintained after controlling for other forms of participation. Results showed that of all gambling activities, EGMs do appear to have the strongest association with problem gambling. Despite having a disproportionately higher level of participation on racing and casino games as compared with other gamblers, problem gamblers are more likely to report regular or weekly participation in EGM gambling and this may be the reason why this activity emerges most strongly as a predictor of problem gambling in multivariate models. This finding is particularly salient, given the very high prevalence of EGM participation, compared to other risky gambling forms. The findings underscore the importance of survey reporting that presents results in a form that can inform policy relevant research relating to the potential impact of different gambling activities.