Cotton fruit from a given location on the plant were assigned to cohorts. The quantity and causes of fruit abscission in each cohort were monitored in three management systems. The proportion of fruit abscission caused by insect pests was largest in the early season cohorts, whereas physical stresses caused increasingly more fruit abortion in later cohorts. Plants tolerate insect-induced fruit abortion by retaining more of the surviving fruit within the same cohort. When a cohort was severely injured, fruit survivorship was increased in subsequent cohorts to compensate for the loss. Fruit production, abscission, and retention were normally distributed, and the patterns were consistent enough to suggest that a cohort technique may be a useful modelling approach.