This study evaluates drought stress effect on winter wheat. Simultaneously, the ability of the HERMES crop growth model to reproduce the process correctly was tested. The field experiment was conducted at Domanínek station in 2014 and 2015, where mobile rain-out shelters were installed on plots of winter wheat (May 2015). Precipitation was reduced in three replications and the findings were compared with results from control plots with ambient precipitation. A precipitation reduction of 93 mm led to reduced growth and decrease in grain yields. The results of this study showed that the model was able to reproduce soil moisture content well and reproduce the drought stress for crop yields of winter wheat to a certain extent. When rain-out shelters were used, real winter wheat yields were reduced by 1.7 t/ha. The model underestimated the yields for the sheltered variant by 0.67 t/ha on average against observed yields and overestimated development of leaf area for both unsheltered and sheltered variants. This overestimation was partly explained by the effect of excluded UV radiation. The outcome of this paper may help to reduce uncertainty within simulated yields of winter wheat under extreme weather conditions through a better understanding of model behavior.