The paper presents the problem of assessment of risk in financial models. This is important from practical point of view since inappropriate use of models in financial markets may cause large losses. The paper describes the sources of model risk and the methods of measuring and handling this risk. Two types of measures are proposed: distribution based measures and sensitivity measures. The remaining part of the paper contains two examples. The first one concerns risk of optimal two-stock selection related to estimation of the correlation coefficient, the second one concerns the risk of option pricing related to estimation of the volatility parameter.