We tested the hypothesis that restricting comparison of the live birth rate following in vitro fertilisation (IVF) treatment in those couples having their first IVF cycle in whom the female is under 35 years of age and has a normal follicle-stimulating hormone level would improve the validity of comparing IVF clinics' success rates. We analysed all cycles performed over a 2-year period in patients who fulfilled these criteria and divided the study population according to the referring primary care trusts: group A (n = 90) were referred from Lambeth, Southwark and Lewisham and group B (n = 134) were referred from Brent and Harrow. There was no significant difference between the two groups with regard to their IVF cycle characteristics. The two groups differed in their ethnicity, cause of infertility, prevalence of uterine fibroids and smoking and alcohol consumption habits. Group A had a significantly lower live birth rate (OR = 0.45, 95% CI 0.21-0.95, P = 0.02) compared with group B. This study confirms the impact of the non-IVF-related patient characteristics on treatment outcome and the poor validity of comparing IVF clinics' success rates based on the sparse data published by national IVF registries.