The association between exposure to ambient air nitrogen dioxide and cough was evaluated in a panel study among 162 children aged 3-6 y. The weekly average nitrogen dioxide exposure was assessed with Palmes-tube measurements in three ways: (1) personally, (2) outside day-care centers, and (3) inside day-care centers. Ambient air nitrogen dioxide concentrations were obtained from the local network that monitored air quality. The parents recorded cough episodes daily in a diary. The risk of cough increased significantly (relative risk = 3.63; 95% confidence interval = 1.41, 9.30) in the highest personal nitrogen dioxide exposure category in winter, and a nonsignificant positive trend was noted for the other assessment groups. In spring, risk increased nonsignificantly in all exposure-assessment groups, except for the fixed-site monitoring assessment. It is important that investigators select an exposure-assessment method sufficiently accurate to reflect the effective pollutant dose in subjects.