The aim of this study was to evaluate objectively whether or not discontinuous albumin gradients enrich the proportion of Y-bearing human sperm. A blinded, collaborative trial design was employed whereby a licensed centre prepared the sperm fractions using licensed procedures, coded the sperm slides and then sent them to an independent laboratory for determination of the X:Y ratio in each sperm fraction using X and Y chromosome-specific probes and double label fluorescence in-situ hybridization (FISH). The identification codes and FISH results were collated by an independent third observer. Two albumin gradient methods which are currently used by licensed centres for male sex pre-selection, protocol 3 and modified protocol 3, were tested. Essentially the same results were obtained for the two methods. Highly motile sperm fractions were recovered from the albumin gradients, and the recoveries of motile spermatozoa (1.3-8.5%) were within the optimal range reported to produce maximal enrichment of Y-bearing spermatozoa. FISH analysis, however, revealed no enrichment for Y-bearing spermatozoa with either method, and the overall X:Y ratios were not significantly different from 1.0. Some samples showed marginal enrichment of Y-bearing spermaotozoa, whereas others showed marginal enrichment of X-bearing spermaotozoa. In conclusion, this collaborative study has demonstrated that the protocol 3 and modified protocol 3 albumin gradient procedures do not enrich Y-bearing spermatozoa. The clinical use of albumin gradients for male sex preselection should be reconsidered in the light of this and other evidence.