Objective To explore the exposure-lag-response relationship between temperature and risk of varicella incidence，and to provide a scientific evidence for early warning and precise prevention and control of varicella epidemic.

Methods Daily varicella cases and daily meteorological data were collected in Minhang District， Shanghai from 2010 to 2022. A distributed lag nonlinear model was used to determine the exposure-lag-response relationship between temperature and risk of varicella incidence. Furthermore， effect of temperature on the incidence risk was determined across different age groups.

Results In 2010‒2022， the total number of notified varicella cases was 26 207 in Minhang District， with the highest incidence in the group aged 3‒14 years （50.35%）. The seasonal pattern of daily varicella cases showed a double peak. The large peak was found in November and December， followed by a smaller peak in May and June. Moreover， the distributed lag nonlinear model showed a unimodal curve in the relationship between temperature and varicella incidence. The RR value reached its maximum peak of 1.90 （95%CI： 1.25‒2.87） at 7 ℃. A reverse U-shape was found in the lag-response curves between temperature and varicella incidence. Furthermore， the effect of temperature on the varicella incidence showed a unimodal pattern in the varicella cases aged 3‒14 years. The RR value reached its peak at 11 ℃ （RR=2.89， 95%CI： 1.33‒6.24）. In contrast， the effect of temperature on the varicella incidence in the cases aged 15 years and above showed a unimodal pattern， with RR value reaching the peak at 5 ℃ （RR=2.14， 95%CI： 1.33‒3.44）.

Conclusion The unimodal curve is found in the relationship between temperature and varicella incidence. Low temperature is associated with increased risk of varicella incidence. Children aged 3‒14 years are more susceptible to the effect of temperature on the varicella incidence.