Thirteen regional climate models are forced with ERA40 reanalysis and compared to high resolution gridded observations for the European region to explore the systematic bias of the models. One motivation for assessing the bias is to calculate model skill scores and to assess whether they can be carried to future climatic conditions. It is important to assess the models by geography because biases can have a consideral geography dependency. This study found a non-linear relationship between model bias and observed precipitaiton and temperature. It was also evident that each model had its own intrinsic bias. The authors noted that regions of very hot temperatures were usually simulated with even warmer temperatures. They also found that dry months were generally simulated too wet and wet months typically simulated too dry. Of the models studied, there was less geographical variation in precipitation biases than for temperature. The study found that model biases have the potential to grow for climate change simulations under global warming conditions. This could lead to an overestimation of projected warming. The authors recommend using a simple bias-correction for individual models. This article is useful for understanding model biases in the context of models forced with reanalysis data. Their methods perhaps represent a better means of identifying model bias as opposed to comparing raw model results to observations.