City Council Chest Clinic, Nairobi, Kenya.To determine to what extent the performance of smear microscopy is responsible for sex differences in notification rates.Three sputum samples from TB suspects were subjected to smear microscopy with Ziehl-Neelsen (ZN) and auramine (FM) staining. Lowenstein-Jensen culture was used as the gold standard.Of 998 suspects, 600 (60%) were men and 398 (40%) women. The odds of detecting culture-positive patients with ZN was lower for women (OR 0.67). By examining the first spot specimen, ZN detected 35% of culture-positive males and 26% of culture-positive females. These proportions increased to respectively 63% and 53% when examining three specimens, and to 79% and 74% when using FM. The sex difference reduced and became non-significant (P = 0.19) when adjusted for HIV; however, the numbers involved for HIV stratification were low.The performance of a diagnostic tool contributes to sex differences in notification rates and influences male/female ratios. Women were less likely to be diagnosed (P = 0.08), and when ZN was used they were less likely to be labelled as smear-positive TB (P < 0.01). The application of more sensitive diagnostic tools such as FM is to the advantage of women.