【Objective】Study the bacterial food poisoning forecasting and early warning system based on meteorological factors. 【Methods】Analyze the characteristics of bacterial food poisoning occurring in Shanghai from 1992 to 2006，build the model for bacterial food poisoning predicting with the probability forecasting method，and make the initial verification of the model with bacterial food poisoning cases occurring in 2007 and 2008. Classify the early warning levels according to the probability of poisoning predicted with the model，and develop the early warning procedures and the corresponding indication term for each early warning level. 【Results】The seasons from May to October are the peak seasons occurring bacterial food poisoning in Shanghai，and the Golden Weeks of May 1 and October 1 holidays，and the period from late June to early July are the three peak periods.

The monthly average number of poisoning accidents is related to the monthly average temperature and relative humidity，and the probability of daily poisoning cases is related to the mean temperature （May-June，September-October） and relative humidity （May-June） on the day of food processing. In this model，the average temperature and relative humidity are selected as the forecast factors，the probability of the bacterial food poisoning accidents occurring in the range of average temperature and relative humidity is calculated according to the historical data. Then the probability of the poisoning accidents on the day is verified according to the forecasts of temperature and humidity. The probability of the poisoning incidents in the Golden Weeks of May Day and National Day is significantly higher than other days，so additional holiday factor is added for correction of the forecast of the probability during the period. Initial verifications show that the forecasting ability of the model for poisoning accidents during the peak period.