A model of constrained utility maximizing behavior is developed to explain how a representative individual allocates his ski days among alternative sites. The physical character­istics of the ski areas and the individual's skiing ability are explicit arguments in the utility function; the budget allocation is given along with the parametric costs to ski (including travel costs, entrance fees, equipment costs, and the opportunity cost of his time). Shares (a site's share being the proportion of ski days that the'"individual spends at that site) are derived and assumed multinomially distributed, a stochastic specification which maintains the inherent properties of the shares. Maximum likelihood estimation confirms the basic hypothesis that costs, ability, and characteristics all are important determinants of the sites' shares. The model explains a large proportion of the skier's allocation of ski days. A multinomial logit model of skier behavior is also developed and maximum likelihood estimates of its parameters are obtained. Examination of the summary statistics from my model and the logit model indicates that my model predicts the skier's choice of sites better than the logit model.