This paper examines the effect of competition on the behaviour of Australian general practitioners. The paper represents a considerable improvement on the methods of previous studies by using a random effects probit model in a multilevel modelling framework to obtain a more robust estimate of the effect of GP density, by including data on GP and practice characteristics and by using data with the actual GP consultation as the unit of observation which are disaggregated by medical condition. This latter characteristic enables us to test the hypothesis that the effect of competition varies across medical conditions. The main hypothesis tested is that GPs in areas of high competition are more likely to recommend a follow-up consultation compared to GPs in areas of low competition. The results suggest that the density of GPs influences the decision to follow up for one out of the four medical conditions analysed. However, there are other issues to be resolved before such results can be confidently interpreted as evidence of the effect of competition and as evidence of supplier-induced demand.