Abstract: Pollen grains of three species of Cupressaceae and five species of Populus were examined in detail by use of a digital photographic technique and MATLAB software. For each pollen species, 50 measurements were taken of maximum, minimum and mean values of area, eccentricity, pole-length and equator-length. P/E and area standard deviations were computed. Variance analysis of the areas of 8 pollen species showed that the method can distinguish the 8 pollen species with reliability of 99%. Cluster analysis of mean eccentricity, pole-length, equator-length and P/E parameter of 5 species of Populus showed that the 5 species can be divided into 3 groups. P.canadensis Moench and P. simonii Carrière form one group; P. tomentosa Carrière and P. davidiana Dode form another group; P. cathayana Rehder stands alone as the third group. This outcome is consistent with scanning electron microscopy results. The method offers a speedy and convenient way for quantitative analysis of pollen morphology, provides more useful data for pollen analysis and improves the accuracy, by about 10 times, than with estimation by the naked eye.