In this paper, we attempt to evaluate whether a film's commercial performance can be forecast. The statistical distribution of film revenues in the UK is examined and found to have unbounded variance. This undermines much of the existing work relating a film's performance to its identifiable attributes within an OLS model. We adopt De Vany and Walls' approach and transform the revenue data into a binary variable and estimate the probability that a film's revenue will exceed a given threshold value; in other words, the probability of a blockbuster. Furthermore, we provide a sensitivity analysis around these threshold values.