Root length density (RLD) distribution information is essential to crop growth and soil water/solute transport predictions and for rational irrigation scheduling and crop management. Unfortunately, measuring RLD accurately in the field is difficult and time-consuming, especially for row crops such as cotton. Field experiments were conducted over two seasons (Exp. 1 and 2 in 2011 and 2012, respectively) to investigate two-dimensional RLD distribution of cotton under film mulched drip irrigation in Xinjiang, China. Different RLD distributions were created by varying row and drip line spacing, irrigation water salinity, irrigation volume and fertilizer application rate. Both distributions of measured RLD and calculated normalized RLD (NRLD), the root length allocation proportions of a plant at different relative depths, were found to decrease with depth, concentrating within the upper soil layers with nearly negligible differences along the horizontal direction. NRLD values in Exp. 2 were pooled together and simply generalized as a vertical one-dimensional decreasing function with just one fitted coefficient of a = 2.06, which was employed to characterize the maximum value of NRLD at the soil surface and its downward decreasing rate. The generalized function was verified to be in good agreement with the measured NRLD distributions of cotton in Exp. 1, and then further improved using all the NRLD data from Exp. 1 and 2. The improved function (with a = 1.96) was shown useful in estimating RLD distribution and simulating soil water flow and salinity transport in a soil–cotton system.