The present study was an initial attempt to calibrate a phenological process-based model of dormancy release with experimental data for apricot flower buds. A large experimental database (88 data points) on dormancy release, concerning several cultivars grown at different geographical sites, was used for the model parameterization. We compared five phenological models. None of them provided accurate prediction of the date of dormancy release at the species level. This inaccuracy appeared to be due to the high variance in dormancy release dates among cultivars. Models fitted for different dormancy release precocity groups provided much more accurate predictions. Parameter estimate analysis of the best model for each cultivar group showed very marked differences in apricot flower bud response to temperature within the species. While in early cultivars dormancy release seemed to be driven by the daily minimum temperature, the daily mean temperature appeared to be the controlling factor in intermediate and late cultivars. Our results show that the apricot dormancy release date cannot be predicted accurately at the species level and that different models should be used for different precocity groups.