In sexual-assault cases, autosomal DNA analysis of gynecological swabs is a challenge, as the presence of a large quantity of female material may prevent detection of the male DNA. A solution to this problem is differential DNA extraction, but there is no established best practice for this. We decided to test the efficacy of a number of different protocols on simulated casework samples. Four difficult samples were sent to the nine Swiss laboratories active in forensic genetics. In each laboratory, staff used their routine protocols to separate the epithelial-cell fraction, enriched with the non-sperm DNA, from the sperm fraction. DNA extracts were then sent to the organizing laboratory for analysis. Estimates of male:female DNA ratio without differential DNA extraction ranged from 1:38 to 1:339, depending on the semen used to prepare the samples. After differential DNA extraction, most of the ratios ranged from 1:12 to 9:1, allowing detection of the male DNA. Compared with direct DNA extraction, cell separation resulted in losses of 94-98% of the male DNA. As expected, more male DNA was generally present in the sperm than in the epithelial-cell fraction. However, for about 30% of the samples, the reverse trend was seen. The recovery of male and female DNA was highly variable, depending on the laboratory involved. An experimental design similar to the one used in this study may be of assistance for local protocol testing and improvement.