The result of empirical testing of forensic DNA match probabilities for Croatia is reported. It is concluded that if consideration is given to relatedness and subpopulation effects the model of Balding and Nichols appears to give very good predictions.