Aims: A range of new differential and confirmation plating media for some non-O157 Shiga toxin producing Escherichia coli (STEC) serotypes (O26, O103, O111, O145) and both sorbitol-positive and -negative O157 were evaluated using artificially contaminated samples. Methods and Results: Dairy products (raw milk, cheese made from pasteurized milk and raw milk), meat (ground beef, fermented meat) and cattle faeces were artificially contaminated using clinical STEC strains. Isolation efficiency was 100%, 82.3%, 88.5%, 65.9%, 64.3% and 15.8%, respectively, for an inoculum size of <= 100 CFU 25 g(-1). The consecutive use of differential and confirmation media limited the incidence of false positive isolates from 0% for raw milk samples, cheese made from pasteurized milk and for fermented meat to 2.1% for cheese made from raw milk, and to 8.9% for ground beef. Conclusions: Data presented in this paper indicated that the efficiency of the applied isolation method was dependent on sample-to-sample variation but not on the inoculum size. Significance and Impact of the Study: Data in this paper indicated that isolation of low levels of non-O157 and sorbitol-positive O157 STEC from food samples is possible.