Studies of associations between the CYP2D6 polymorphism and susceptibility to specific diseases, particularly lung cancer and Parkinsonism, have produced conflicting results with respect to an under or overrepresentation of poor metabolizers. Accordingly, we have re-evaluated this primary research (18 studies on lung cancer and 18 on Parkinsonism) using meta-analysis. For lung cancer, the median odds ratio (OR) was 0.69 (95% confidence interval (CI) 0.52-0.90), which differed significantly from unity (P < 0.007). A trail comprising 3000 patient and an equal number of control individuals would be required to demonstrate that this observation had arisen purely by chance (i.e. OR = 1). For Parkinson's disease, the analysis gave an OR of 1.32 (95% CI 0.98-1.78), which was of borderline statistical significance (P < 0.074). If the only individual study that was statistically significant was excluded, the P-value increased greatly to 0.489. A study of at least 500 patients and an equal number of control individuals giving the same value as the current mean OR of 1.32 would be required to make the overall analysis statistically significant. In summary, poor metabolizers with respect to CYP2D6 show a small decrease in susceptibility to lung cancer compared with extensive metabolizers and its is hard to justify further studies. The relationship between the CYP2D6 polymorphism and lung cancer, as a determinant of individual susceptibility, is not appreciable (OR = 0.69) compared with that between smoking and lung cancer (OR > 11). Nevertheless, the epidemiological impact on the number of poor metabolizers who are protected from lung cancer may be considerable. With regard to Parkinson's disease, additional well designed studies may allow a definitive conclusion, although any risk for poor metabolizers is likely to be small and therefore of questionable clinical significance. An important lesson from the current review of studies is that much time, effort, expense and patient inconvenience might have been avoid if more attention had been paid to appropriate study design particularly in the selection of control groups.