Mutagenicity with Salmonella strain TA 100 can be determined by an island test in which only an isolated portion of a pour plate is spotted with a sample. The magnitude of the ratio of the numbers of revertant colonies on the treated and untreated parts of the plate reflects the potency of the mutagen. Five of six chemicals tested yielded statistically significant and generally linear dose-response curves. Minimum detectable mutagenic doses for four compounds calculated from the dose-response curves showed that the island test generally required less material for detection of mutagenicity than the liquid pre-incubation procedure. The island method was primarily designed to test the mutagenicity of aqueous extracts of human stool samples. Dose-responses were obtained for six such samples, and in two samples the amount of material needed for a positive response was significantly less than that required in the liquid pre-incubation method.