Low activity of arylamine N-acetyltransferase 2 (slow NAT2) was consistently associated with urinary bladder cancer risk. The increased cancer risk attributable to slow NAT2 was more significant when taking gene-environment interactions and gene-gene interactions into account. In urinary bladder, slow NAT2 was no risk factor in subjects who never smoked but became increasingly relevant with increasing lifetime dose of tobacco smoke expressed by an odds ratio of 2.7 for slow NAT2 in extensive smokers. The functional impact of some arylamine N-acetyltransferase 1 variants is controversial. It was published that the NAT1 allele 10 was associated with high enzyme activity and that there was an overrepresentation of carriers of NAT1*10 in bladder and colon cancer, but we could only detect a moderately elevated activity of NAT1*10 and an underrepresentation of fast NAT1 alleles in bladder cancer. Recently, a C/A-polymorphism in intron 1 of cytochrome P450 1A2 was associated with high inducibility and persons with this high inducibility variant were overrepresented in bladder cancer, but only if they were smokers or if they had slow NAT2 genotypes. Numerous studies have shown that glutathione S-transferase M1 deficiency (GSTM1*0/0) increases the risk for lung and bladder cancer but the overall risk attributable to GSTM1*0/0 was only around 1.3 according to meta-analyses. The GSTM1*0/0 genotype appears to be the best established metabolic susceptibility factor. Several independent experimental approaches showed that GSTM1 decreases mutagenicity of reactive epoxides and it was shown that carriers of GSTM1*0/0 were at increased risk for several types of cancer and other diseases. There are also studies which showed no effects of GSTM1, a result which is compatible with the assumption that GSTM1*0/0 is a susceptibility factor of moderate strength. GSTM1*0/0 may, however, become a dominant risk factor in certain gene-gene combinations such as the combination with highly active CYP1A1 gene variants or in combination with specific types of exposure. Specific precautions have to be taken in the design of molecular epidemiological studies on risk factors with moderate strength; some requirements for high quality molecular epidemiological studies will be discussed in this article. Molecular epidemiology is an increasingly powerful approach to understand carcinogenesis and may be used in the future to individualize cancer prevention strategies.