To determine whether gambling participation falls into skill and chance-based categories and, if so, to determine the socio-demographic characteristics associated with these different categories. A cross-sectional analysis of all respondents to the 2005 Northern Territory Gambling Prevalence Survey who gambled in the 12 months prior to the survey. Factor analysis was employed to determine whether a chance versus skill-based dichotomy described the structure of gambling participation. Gambler preference groups were constructed using the median of rotated factor scores. Multinomial logit regression was then used to determine independent associations between explanatory variables and categories of gambler preferences. The skill and chance-based dichotomy did describe player preferences for the sample of adult gamblers in the Northern Territory, Australia. Gender, age, household income, household structure and the geographic location (access to gambling opportunities) of respondents were all associated with different degrees of participation in skill and chance based gambling activities. Notably, respondents 35 years and over were significantly over-represented in the low-skill/high-chance participation group, and under-represented in the high-skill/low-chance group. It is clear that the term gambling is a confounding rubric that hides differences both in the type of activity and the socio-demographic profiles of participants. An examination of the latter raises important questions about the role of chance in later life, as well as the role of self-determination in gambling for other groups, particularly younger men.