 This study compared the performance of various models used to predict runoff hydrographs and ungaged catchments. It found that human catchments were more accurately predicted than arid ones and that larger catchments were also more accurately predicted. Additionally, it found that the choice of regionalization method had an impact on the accuracy of the prediction with spatial proximity and geostatistical methods performing better than regression or average based methods. Finally, it showed that the number of regionalized parameters did not necessarily correlate with improved prediction accuracy.