In 1996 and 1997, various intensities of prebloom square removal were applied to three cultivars of cotton grown in Mississippi. With the exception of one cultivar in 1997, all cultivars were B. thuringiensis (Bt)-transgenic cotton. At harvest, the number of bolls and seed cotton weight was recorded for all plants in each square removal treatment. All cultivars responded similarly to square loss. A yield increase (overcompensation) was observed in the treatment where all squares were removed from the plant one week after squaring began. Only the treatment where all squares were removed before bloom significantly reduced yield and caused a large (>7 d) delay in crop maturation. Otherwise, moderate levels of square removal (approximately 20-50% of prebloom squares) had little impact on overall lint production. However, the patterns of cotton production on the plants were significantly influenced by the square removal treatments. The removal of relatively more or larger squares increased seed cotton production in late-season fruiting cohorts and on 'vegetative' branches. Compensation for square loss occurred by increasing the relative number and weight of bolls produced subsequent to early-season square removal. Typically, early-season square loss increased the value of later-season fruiting cohorts, especially the midseason cohorts and bolls on vegetative branches. The implications of prebloom square loss, including the compensatory ability of the cotton plant, on insect management are discussed.