Background and aims: It is well understood that engagement with some forms of gambling, like EGMs, is riskier than other forms. However, while reports of associations are common, few studies have attempted to evaluate and compare the relative risk of all available forms, and none have estimated the relative contribution of each form to the total burden of gambling problems (GP) in a population. Methods: Using an aggregated dataset of national and state-based prevalence studies in Australia (N 5 71,103), we estimated prevalence and unique effects of frequency of engagement on each form on GP. Two alternative numerical methods were then applied to infer the relative contribution of each form to the total amount of GP. Results: EGMs are responsible for 51%-57% of gambling problems in Australia, and 90% of gambling problems are attributable to EGMs, casino, race, and sports betting. Casino table games and EGMs are equally risky at the individual level, but the former contribute far less to problems due to low participation. Bingo and lottery play show no statistically detectable risk for GP. Discussion and conclusion: The results illustrate which forms present the greatest population burden and illuminate the reasons why. EGMs have an outsized impact. EGM uniquely combines high risk conditional on play, with a high participation rate and a high frequency of play among participants. This is in contrast to risky but less commonly played casino games, and prevalent but non-risky forms like lotteries. We conclude that EGM regulation should be a primary focus of policy action in Australia. More innovative policy ideas relating to EGMs should be tested due to the disproportionate impact of this product type.  Kulcsszavak: gambling, gambling problems, gambling harm, electronic gaming machines, gambling policy, harm minimisation