Fruit set was studied over three years in 11 almond (Prunus amygdalus Batsch) selections of the Zaragoza breeding programme, previously selected for their fruit and kernel traits and self-compatibility, following five different pollination treatments: open pollination, open pollination of a reduced number of flowers, artificial self- and cross-pollination after emasculation, and autogamy in bagged branches. Significant differences were observed in all cases: among selections, among years and among treatments. There was also a significant interactions for all cases. Cross-pollinations as a whole produced better sets than self-pollinations, but not in all cases, showing differences in the ability to set self fruit among the selections. Sets in the bagged branches were the lowest, showing the differences in autogamy ability of each selection. Year effects were observed, but not affecting all the treatments and selections at the same level, thus stressing the need for autogamy evaluation during more than one year.