Interest in candidate genes in psychiatry has been revived by the success of that approach in Alzheimer's disease as well as by discouragement over the results to date of searches for linkage. If mutations in candidate genes could be detected directly many of the complexities of searching for them indirectly through linkage could be avoided. As attractive as this approach may sound, mutation searches with candidate genes will introduce a new set of difficulties: the large number and low a priori probability of potential candidate genes virtually guarantees that most positive findings will be false positives. Consequently, if candidate genes are to be systematically studied, procedures need to be established for minimizing and efficiently disconfirming false positive results.