This paper describes the methods used and presents an analysis of the results obtained from three years of study, to determine the separate and inter-related effects of various agents such as gravity, wind and insects, in the pollination of flowers of Coffea arabica L. Observations were made and data obtained from several thousands of normal and castrated flowers that were maintained under natural and controlled conditions. It has been found that the importance of gravity, wind, and insects in pollination of the flowers may vary appreciably in relation to local environmental influences. The data obtained, however, indicate certain trends that are of definitive interest. Based on the total number of ovules, it was found that in 1228 normal flowers observed, 62 percent produced seed. This value is believed to represent in general the percentage of fertilization that might be expected to occur naturally. In tests designed to exclude the influence of wind, insects, and gravity, it was found that an average of 24 percent fertilization (within flowers) occurred. In measuring the effects of the combined agents of wind, insects and gravity it was found that 18.5 - 32.7 percent fertilization occurred as a result of self-pollination and 4.1 to 5.2 percent was due to cross-pollination. Analysis of the data also show in all except one case, the percentage of fertilization resulting from self-pollination was higher than that from cross-pollination. These findings show the importance of self-pollination in Coffea arabica. The use of castrated flowers was particulary helpful in determining the maximum effect of each of the pollinating agents studied under isolated controlled conditions. The general analysis of all data indicates that in the case of self-pollinated normal flowers the influence of wind and insects are about equal and that the effect of gravity is relatively less and likely to be of variable importance. In the case of factors affecting cross-pollination of normal flowers, wind seemed to have the most important influence. Based on a study of several different samples with large numbers of seed harvested from normal flowers, it. was also found that the percentage of seed resulting from cross-pollination was 7.3 - 9.0%. This percentage range of 7.3 - 9.0 is comparable to that of 4.1 - 5.2 shown above, the latter percentage range being based on total ovules and the former on the total seed harvested. Again these data emphasize the relatively limited occurence of cross-pollination in C. arabica.