The application of the polymerase chain reaction (PCR) for detection of Clostridium botulinum types A, B and E in foods, environmental and clinical samples was evaluated and compared to the mouse bioassay. Samples inoculated with 10, 100 and 1000 spores of Cl. botulinum types A and B included pasteurized milk, UHT milk, infant formula, infant faeces, meat juice, canned tuna, mushrooms, blood sausage and soil. Clostridium botulinum type E spores were inoculated into fish eggs, canned tuna, picked herring, raw fish and soil at similar levels. Spores were added to 2.5 g of each sample with the exception of soil which was inoculated in 10 g samples. The presence of Cl. botulinum in sample enrichments was determined by both PCR and the bioassay. An overall correlation of 95.6% was observed between PCR results and the mouse bioassay. Of the total of 114 samples tested there was disparity between the mouse bioassay and the PCR in three samples of soil inoculated with 100 type A or E spores and 10 type B spores per 10 g, respectively, and two samples of infant faeces inoculated with 10 type A or B spores per 2.5 g. All of these samples gave negative animal results and positive PCR results.