Gossym-Comax was evaluated in field experiments in Alabama during 1987 and 1988. Two nonirrigated experiments were conducted in 1987 and two nonirrigated and three on-farm irrigated experiments were conducted in 1988. In both years, Gossym-Comax accurately predicted first square and first bloom for cotton. With a late season drought in 1987, Gossym-Comax was found to recommend too much nitrogen (N) fertilizer and was one to two weeks late in predicting defoliation. Many changes were made in Gossym-Comax for the 1988 season and data from 1987 were again evaluated using the 1988 version of Gossym-Comax. Results were very encouraging because simulated nitrogen fertilizer rates and defoliation recommendations agreed well with the 1987 data. Projected cotton yields, however, were 12 and 50 percent greater than actual harvested yields. In 1988 an early drought severely stunted cotton growth and defoliation treatments were abandoned. With this early drought, Gossym-Comax under-estimated the nonirrigated cotton yields. Gossym-Comax predicted cotton lint yields of between 81 and 155 lb/A whereas actual harvested yields averaged 396 lb/A. Under irrigation, Gossym-Comax more accurately predicted cotton development. Using an "early" variety file and an experimental PIX program, Gossym-Comax accurately predicted N fertilizer rate, defoliation dates, and lint yields in the three on-farm tests. Although it did not affect yields, Gossym-Comax did under-estimate cotton plant height in the irrigated cotton, especially with the experimental PIX program.