Credibility theory is an experience rating technique in insurance used to combine an estimate of the expected claims of a contract with the estimate of the expected claims of a portfolio of similar contracts. However, the credibility estimate remains sensitive to large (outlying) claims. In this thesis, robustification of some classical credibility models are presented via robust Kalman filtering. Credibility theory has been shown to be a special case of the Kalman filter (De Jong and Zehnwirth, 1983), thus existing research on the robustification of the Kalman filter, for example, Cipra and Romera (1991), can be applied to robustifying Kalman filter credibility models (Kremer, 1994). After describing in some detail the classical and robust models of credibility, we present an implementation of a robust Kalman filter credibility model and apply it to Hachemeister's dataset (Hachemeister, 1975).