The decision to terminate the irrigation of cotton (Gossypium hirsutum L.) is complicated by interactions and uncertainties related to lint yield and quality, water costs, and market factors. High, medium, and low values for the cost of water, lint price, and quality discount/premium schedule for High Volume Instrument (HVI) quality factors were applied to lint yield and quality differentials realized from irrigation termination experiments conducted in central Arizona for the crop years of 1991 and 1992, 1994 through 1997, and 2000. Deviations in lint yield and quality for the first irrigation termination treatment versus the subsequent second and third irrigation termination treatments are the agronomic basis of this study. Irrigation termination dates were defined by heat units after planting (31/12.8º C or 86/55º F) to place the crop progression of different experimental sites and years on a more equal basis than using calendar dates. Classification and regression tree analysis was used to quantify the data. The relative ranking of results from this procedure in which the most important variable is normalized on 100 were as follows: cultivar (100), additional heat units after the first irrigation termination treatment (94), yield of first irrigation termination treatment trial (93), year of the field experiment or crop year (83), micronaire associated with the first irrigation termination treatment (68), heat units after planting for the first irrigation termination treatment (67), lint price (5), water cost (2), and the quality discount/premium year (0.09). In general, the season needs to be extended at least 330 to 360 heat units Centigrade (600 to 650 heat units Fahrenheit) to yield a profitable return for cultivars with potential to produce a top-crop. Also, a crop that has a yield less than 1533 kg ha-1 (1368 lb ac-1) at the first irrigation termination treatment is more likely to have the potential for producing a profitable top-crop than a crop that has already set a fruit boll load greater than this level.