Abstract: Drought is one of the meteorological hazards that severely affects winter wheat production in Hebei Province. Accurate monitoring and prediction of drought occurrence provides the scientific basis for hazard control decision-making. This paper analyzed drought conditions in the winter wheat production belt of Nangong County, South Hebei Province. The study used observed agro-meteorological data and regular meteorological data for 1991~2007 to establish a drought index and prediction model for winter wheat. The water sensitive coefficients of winter wheat during re-greening to jointing, jointing to heading, heading to milky maturity and milky maturity to grain maturity stages were calculated using the Jensan model. Then percent yield reduction and relative evapotranspiration at each growth stage were used as index value to determine light drought, moderate drought, heavy drought and severe drought. A simulation test for 12 years (1991~2005) with 2006 and 2007 as evaluation periods was conducted. The results showed that the established index values for drought degree in different growth periods objectively reflected the active drought degree in the region. The criteria for drought degree took into account of the sensitivity of winter wheat at different growth stages. Regression models were used to predict drought in the four growth stages of winter wheat. The established drought prediction model results were significant at P=0.05. The correct rate of model simulating was 70.8%, correct rate of prediction was 75.0%, and the average correct rate was 71.4%. Assuming that drought was classified as level-one drought (light drought) and level-two drought (medium, heavy and severe droughts), then the model simulating correct rate was 81.3%, correct rate of prediction was 75.0%, and average correct rate was 80.4%. In summary, model calculations and predictions were in good agreement with observed data. Thus the prediction model had practical application in early warning and control of the impact of drought in winter wheat production.