Little is known about the association between air temperature and causes-specific respiratory diseases (RD), especially in northwest China. A time-stratified case crossover design with a distributed lag nonlinear model (DLNM) was conducted to assess the nonlinear and delayed ef-fects of temperature on total and cause-specific outpatient visits, with analyses stratified by gender and age. The cumulative effects of temperature were irregular M-shaped curves for total and upper respiratory tract infection (URTI), with inverted U-shaped curve for pneumonia, bronchitis and chronic obstructive pulmonary disease (COPD). Positive cold and heat effects were observed for URTI, pneumonia, bronchitis and COPD, bronchitis was most vulnerable to cold but pneumonia was more affected to heat. Heat effects were immediate whereas cold effects were delayed and lasted longer. The magnitude of temperature effects varies greatly by age, gender, and disease. Notably, the cold effect was greater for children aged 0-14 than that of other age groups. This study suggested that both cold and hot temperatures exposure could increase all-cause and cause-specific respiratory outpatient visits in Lanzhou, China. The harmful effect and duration of cold were greater than that of heat, and children aged 0-14 were more sensitive to cold. Protection against extreme temperatures should be strengthened.