Nitrogen sources and application methods in no-till cotton production have received little attention by researchers. The objectives of this study were to (i) determine if no-till cotton yields respond differently to injected and broadcast N, (ii) determine if no-till cotton yields respond differently across locations, and (iii) identify economically optimal N rates, yields, and net revenues. Yields were obtained for broadcast and injected N rates of 0, 34, 67, 101, and 134 kg ha-1 on Loring (killed native winter vegetation), Lexington (killed winter wheat), and Memphis (corn stover) silt loam soils at Milan, TN (1994 through 1997), Jackson, TN (1996 and 1997), and Ames Plantation, TN (1996 and 1997), respectively. Quadratic yield response functions were estimated and tested for significant differences among application methods and locations. The broadcast and injected functions were not significantly different, but significant differences were found across locations. Economically optimal N rates were 112, 92, and 107 kg ha-1 at Milan, Jackson, and Ames Plantation. These rates were 24, 2, and 19% higher than the 90 kg ha-1 maximum currently recommended for cotton production in Tennessee. Optimal yields were 1,291, 1,510, and 1,117 kg ha-1, and optimal net revenues above N costs were $1,955.80, $2,189.72, and $1,600.73 ha-1, respectively. Only 3% ($9.80 ha-1) of the net revenue difference between Milan and Jackson came from the difference in N cost, while this percentage was only 1% ($7.35 ha-1) when Ames Plantation and Jackson were compared. The remainder of the net revenue difference came from differences in optimal lint yields.