Gene flow by pollen between trees is essential for nut set in commercial almond orchards, due to the self-incompatibility of almond cultivars used. A study of gene flow has been carried out in an orchard composed of single rows of a "pollinating" cultivar between every double row of the most commercially desirable cultivar, Nonpareil. This "two-to-one" planting pattern was repeated throughout the orchard, and several "pollinating" cultivars were used in various parts of the orchard in an attempt to provide flowers for cross-pollination with Nonpareil at all stages of flowering of the latter. Using isozyme markers GPI-2, LAP-1, AAT-1, PGM-1, and PGM-2 and three newly-defined isozyme markers for almond - IDH, G6PD, and SDH - it has been shown that the gene flow resulting in nut set is quite restricted, taking place most strongly between neighboring halves of cross-compatibile pairs of trees. Even that half of a tree facing away from the "pollinating" tree has significantly less gene flow to it, while the next tree further on has few nuts set by fertilization from the "pollinating" tree under consideration. This result is surprising considering the comparatively large distances that the honeybee brought into the orchard in large numbers must travel within the orchard. To explain this apparent paradox and the observation that in most cases only a small proportion (<20%) of flowers set nuts, it is suggested that the honeybee predominantly visits only one cultivar, flying along the row of the cultivar to do so, and that cross-pollination results from accidental or rare visits involving two or more compatible cultivars.