In this study, an attempt was made at predicting the values of selected reproductive parameters in Harnai sheep using different data mining algorithms (artificial neural networks - ANN, classification and regression trees - CART, chi-square automatic interaction detector - CHAID and multivariate adaptive regression splines - MARS) and indicating the most influential predictors of these traits. A total of 382 reproduction records including three predictors (month of lambing - MOL, age at first lambing - AFL and lambing weight - LW) and seven dependent (output) variables (services per conception - SPC, service period - SP, lambing interval - LI, twinning rate - TR, gestation length - GL, breeding efficiency - BE and fertility rate - FR) were used. A 10-fold cross-validation was applied to train and evaluate the models. The highest correlation coefficients (r) were found for LI (0.18 - 0.29; P≤0.001), GL (0.05 - 0.21; P≤0.001 to P>0.05) and FR (0.11 - 0.26; P≤0.001 to P≤0.05). For the remaining output variables, it was usually lower than 0.10. The smallest values of SDratio (0.96 - 1.06) were found for LI, GL and FR. For the rest of the output variables, it was usually above 1.00. The measures of predictor importance to ANN, CART, CHAID and MARS were generally low. In conclusion, the applied method of reproductive parameters prediction was rather ineffective, indicating that more powerful input variables are required to obtain better prediction results.