The behavior of the betting public was examined at over 8,000 horse races at twelve major race tracks. It was hypothesized that the public was trying to pick winners, but also to apply the laws of successive interval scaling: ranking and sorting horses into categories; assigning probabilities of finishing first, second, or third, as well as seventh or eighth; and simultaneously providing payoffs. All horses having the same proportion of win money bet on them were identified as a set, and were examined with regard to their order of finish; the resulting matrix was consistent with the Successive Intervals Scaling model. Additional analyses were conducted for races containing seven to ten horses. Data included scale values of the order of finish, numbers for proportion of money bet (indicating subjective probability), and a set of rules for manipulating those numbers as implied by the laws of categorical judgment. Results confirmed the appropriateness of this model. Results also indicated that the ability of the public in scaling horses was independent of the number in the race. The objective probability of a horse winning was computed, and was highly correlated with the subjective probability. Analysis of the money received versus the money that should have been received (the utility of the money), was constant even if the number of horses changed. Paper presented at Eastern Psychological Association, September 1962.