The genotoxicity of river water samples was evaluated by the Salmonella mutagenicity assay and by the microscreen phage-induction assay. Different processes of sample treatment were compared using the following assays: different volumes of a non-concentrated sample (direct method); concentrated sample fractionated into portions with acid, basic and neutral activity (liquid-liquid extraction method); sample submitted to extraction of volatile substances (volatile extraction method). Samples that were positive to the Salmonella assay by the direct concentration method lost this activity after liquid-liquid extraction. This difference was related to the loss of substances that volatilize during the extraction process. The study of volatile product concentrates confirmed the role of these compounds in inducing activity present in some samples. The microscreen phage-induction assay proved to be a good screening assay for genotoxic compounds present in small concentration in environmental samples. We conclude that, whenever possible, samples should be treated by the direct method in different volumes to prevent the loss of genotoxic substances.