Previous research using the Gambling Outcome Expectancies Scale (GOES; Flack and Morris in J Gambl Stud, 2015. doi: 10.1007/s10899-014-9484-z ) revealed the instrument has excellent psychometric properties and differentially predicts gambling frequency and problem gambling scores. However, like the existing gambling motivation scales, the GOES psychometric properties and predictive utility have not been tested outside of cross sectional studies. The current study used a prospective survey design to redress this issue. Eight hundred and ninety-three participants, drawn from the general community, completed the second wave of the gambling survey. Temporal invariance testing revealed the GOES was reliable. Furthermore, the ability of the GOES to predict gambling behaviour using baseline and concurrent measures of gambling outcome expectancies was demonstrated. Specifically, consistent with the Wave 1 results, the gambling outcome expectancies that reflect diverse reasons for gambling (e.g., social, escape, and money) preferentially predicted gambling frequency whereas the narrower range of emotion focused reasons (e.g., excitement, escape, and ego enhancement) predicted gambling problems. Considered in light of the Wave 1 findings, these results underscore the need for gambling harm minimisation initiatives to take into account the emotion-oriented reasons for gambling.