Three statistical methods - multiple and linear regression, and analysis of variance - were used to interpret the data in this study. The results of a preliminary study indicated that there were several major sources of variation in the non-returns of bulls. These were seasonal differences, possible breed differences, and differences between bulls within a breed. In addition, there were differences in breeds in different seasons. Upon consideration of these data a regression study of 33 bulls was undertaken to ascertain the degree of relationship between the changes in the early and late non-returns to first services. Statistically, it was found that when both the 60-90 day and 30-60 day non-returns are available, the latter are of little practical value in predicting the changes in the 150-180 day non-returns. When only the 30-60 day figures are available they may be used with reasonable accuracy to “estimate” the final changes in the 150-180 day non-returns. An analysis of variance of the errors of estimate from the regression studies indicated that there are several assignable causes of variation in non-return estimates. These include differences between breeds as to the effect of season on reproductive efficiency, differences between bulls within a breed, and a significant difference in seasonal effect in different stations. The use of non-return data in estimating the reproductive efficiency of a bull is of value only when all factors that affect this reproductive efficiency are considered. This investigation shows the significance of some of the various factors to be considered. / Master of Science