This article uses the bounds testing procedure to cointegration, within an autoregressive distributed lag framework to estimate the determinants of attendance at the Melbourne Cup from its inception from 1861 to 2002. Following the literature on the demand for professional team sports, attendance is specified as a function of economic, demographic and race-specific factors. The main findings are that real income and population size are the major determinants of attendance in the long run, while in the short run the weather is the most important factor explaining attendance.