Objective: To compare the pregnancy rates (PRs) after intrauterine insemination (IUI) with frozen-thawed donor sperm. Sperm were isolated after thawing using three different sperm preparation techniques: simple washing, Percoll density gradient, and Sephadex columns (SpermPrep Column; Fertility Technologies, Natick, MA). Design: Women (n = 98) were randomized upon entry into the program into one of three different sperm preparation methods. The same sperm preparation technique was used for the woman during subsequent cycles, if pregnancy did not occur. The study was stopped when greater than or equal to 75 treatment cycles for each group were completed for analysis of PRs per treatment cycle. Setting: All patients were treated at our private care center at the University of Texas Southwestern Medical Center. Patients: Patients entering this study were spontaneously ovulating women undergoing IUI with frozen donor sperm. Main Outcome Measure: Pregnancy rate per cycle of timed IUI. Results: After 260 cycles of insemination, the PR per cycle was 19.1% with simple washing, 16.9% with Sephadex columns, and 11.4% with Percoll density gradient. Although these results were not statistically different, Percoll density gradient had a 40% lower PR per treatment cycle compared with simple washing. However, Percoll density gradient preparation of sperm resulted in a statistically significant increase in the motility, curvilinear velocity, straight line velocity, and the number of normal heads compared with the other two treatments. Conclusion: Although Percoll density gradient separation of sperm results in a population of cells that is more motile and morphologically normal, this did not result in subsequent cycle fecundity. These data suggest that the reliance on the averaged values of motility, curvilinear velocity, straight line velocity and morphology may have little predictive value of the potential fertility of frozen-thawed sperm.