Foliar-applied KNO3 on cotton is becoming a widespread practice in parts of the Southeastern cotton belt. However, predicting the probability of a yield response to this practice is uncertain. In 1992 and 1993, long-term soil fertility experiments at 5 Alabama locations were used to measure cotton yield response to foliar-applied KNO3 and urea on plots where residual soil K variables exist. Foliar KNO3 or urea was applied to split plots at a rate of 4.4 pounds K2O per acre and 1.3 pounds N per acre 4 times, 7 to 10 days apart, beginning the first week after bloom initiation. Cotton at all locations in 1992 and 3 locations in 1993 responded to increasing levels of residual, plow-layer K. However, foliar KNO3 significantly increased yields at only 2 locations in 1992 and at no locations in 1993. Yield response to foliar-applied K does not appear to be dependent upon residual, plow-layer, soil-test K levels or to potential yield. Other soil and physiological factors may influence the predictability of cotton yield response to foliar-applied K.