the present study may not have been large enough to exclude involvement of a polymorphism with weak effect. Furthermore, in our study sample, to detect the difference between 93/196 and 103/196 with power 0.5, the minimum sample size should be 778. Similarly, to detect the difference between 237/461 and 224/461 in the sample from three Chinese studies with power 0.5, the minimum sample size should be 2487. If a higher power is desired, say, 0.8, then the required minimum sample size would be much greater than those two figures above.In this study, we only tested COMT Val108/158Met polymorphism in a family-based association study. There is some evidence that the COMT haplotype has additional predictive power for COMT expression. 8,9 In support of this suggestion, Shifman et al. 10 found schizophrenia linkage only to a Val158-containing haplotype. Thus, further research into genotyping more markers and using a haplotype-based approach is needed to definitively characterise the role of COMT gene in the pathogenesis of schizophrenia.

No abstract