There is now overwhelming evidence for a role of genetic factors in susceptibility to Parkinson's disease. This review concentrates only on the evidence from candidate association studies because the role of linkage analysis and the successes of positional cloning are dealt with by others in this supplement. Candidate gene studies are a useful and powerful adjunct to the armamentarium of the researcher, but the robustness of the data is difficult to achieve. This is illustrated by the large literature on this subject, only a relatively small proportion of which has been reproduced. A recent study by our group is used to illustrate some of these difficulties, and the possible ways around some of the inherent biases in most of the studies are also discussed.