The relation of respiratory illness to nitrogen dioxide (NO2), temperature and relative humidity in homes with a gas cooker was investigated in five and six year old children living in an urban area of the northern England. NO2 was measured for one week in the child's bedroom and living room of each home, and temperature and relative humidity were measured in the bedroom only. Information on respiratory conditions experienced by the child and characteristics of the home was collected in a self-administered questionnaire completed by the child's mother. Access was gained to 183 homes (54.3% of 337) where only gas was used for cooking. Complete information was obtained for 179 (93.7%) of 191 children who lived in the 183 homes. After allowing for the effects of age, sex, social class, and number of cigarette smokers in the home and temperature or relative humidity, no statistically significant relation was found between the prevalence of having one or more respiratory conditions and weekly average levels of NO2 in the bedroom (range 4.7 to 160.8 ppb) or living room (range 9.0 to 292.2 ppb). However, the prevalence of having one or more respiratory conditions tended to be highest in homes with high levels of NO2 and lowest in homes with low levels, consistent with earlier findings. A significant positive association was found between the prevalence of respiratory conditions and relative humidity (p less than .05). A harmful effect on health from NO2 cannot be totally dismissed but if it exists it is weak and difficult to detect in small samples of children.