Recent advances in information technologies are generating a growth in the amount of available biomedical data. In this paper, we studied the impact sample size may have on the categorisation of a continuous predictor variable in a logistic regression setting. Two different approaches to categorise predictor variables were compared.