Based on the outpatient and inpatient data of heatstroke in Tianjin from 2016 to 2020, a comparative analysis of the relationship between meteorological factors and the heatstroke outpatient rate is conducted using the Generalized Additive Modeland Distributed Lag Nonlinear Model. By introducing the human heat balance model and predicted mean vote (PMV), a localized heatstroke meteorological risk warning index is established. The results show that the numbers of outpatient and inpatient cases of heatstroke in Tianjin are concentrated from late June to early August each year, with 84% of the peak heatstroke days occurring in 6 continuous weather processes over the 5 years. The occurrence of heatstroke is most correlated 〖JP〗with the meteorological conditions of that day and the previous day. There is a significant increase in heatstroke cases when the maximum temperature exceeds 35℃. Males are more susceptible to heatstroke than females, and the outpatient rate of the elderly is significantly higher than the general population. The heatstroke outpatient rate is positively correlated to average temperature, maximum temperature, relative humidity, and solar radiation intensity, with the strongest correlation with average temperature but a negative correlation with wind speed. Introducing the human heat balance model, PMV shows a higher correlation with the heatstroke outpatient rate than any single meteorological factor, indicating PMV has a clear advantage in evaluating heatstroke meteorological risk. A forecasting equation with PMV as the key indicator is developed.