While there are many cross-sectional studies that provide data on gambling behaviour and the characteristics of those who gamble, there are few large-scale population based longitudinal studies of gambling. This is a serious limitation. It is increasingly being recognised that longitudinal data is required to properly understand gambling behaviour, trajectories, risks and consequences. This paper makes use of a large-scale longitudinal survey that includes questions on gambling behaviour in Australia: the Household, Income and Labour Dynamics in Australia (HILDA) survey. Although the HILDA survey currently provides data on gambling at a single point in time in 2015, there are data on the individuals, in the majority of cases, back to 2001. This paper analyses the trajectories of a range of social, health and economic outcomes according to self-reported risk of experiencing gambling related problems. The analysis clearly shows that problem gamblers experience significantly worse outcomes than those without gambling problems, and poor outcomes go back a number of years. In a number of cases, outcomes are becoming progressively poorer, which may suggest either increasingly risky gambling behaviour or the cumulative effects of a sustained period of gambling problems. The long-term nature of the poor health and social outcomes experienced by problem gamblers strongly suggests that programs designed to address gambling problems will have to be cognisant of, and address these underlying issues. Effective treatment is thus likely to be long-term and challenging. It also suggests that restricting access to gambling via mechanisms such as pre-commitment or self-exclusion will be effective.