The study involved a sample of 374 students (mostly men) who attended a large eastern university in the United States from 1995 through 1997. All participants completed the South Oaks Gambling Screen (SOGS) to indicate gambling behavior over the past 6 months, as well as the GABS. The 35 GABS items capture a wide range of cognitive biases, irrational beliefs, and positive attitudes toward gambling. Compared to adults in the general population of the university's State, the students gambled at high rates, with 19.5 percent gambling at least once a week, 47.3 percent gambling at least once a month, and 33.2 percent gambling less than once a month. Using techniques based in modern test theory (e.g., Rasch model), the study identified a set of 10 GABS items that reliably rank-ordered students in relation to their positive attitudes and beliefs about gambling. These 10 items accounted for variance in gambling behavior beyond that provided by a more standard consequences-based assessment measure. The 10 items pertain to feeling confident when gambling, knowing when one is on a gambling "streak," believing that betting strategies contribute to wins in any game, having an intuition about when good luck is coming, the identification of streaks as critical in successful gambling, pressing bets following recent wins, believing that luck will change following losses, believing that some people bring bad luck to other people, the experience of being "really alive" when gambling, and believing that the excitement of making a big bet is a key life experience. The study concludes that the GABS' strong psychometric properties make it a promising tool for assessing college students' risk for gambling. 2 figures, 1 table, and 20 references