Empirical works in the Schumpeterian tradition have produced conflicting results regarding the relationship between firm size and R&D intensity. This paper makes a distinction between the decision of the firm to start an R&D unit (a zero-one decision), and a second decision regarding the amount to be spent on R&D. A Probit model is developed to analyse the first decision for a sample that includes both R&D and non-R&D firms, and a linear regression model is employed for a sample consisting of only R&D firms to analyse the second decision. This approach produces better results.