During a survey of foal diarrhoea between 1991 and 1994, Clostridium perfringens was significantly associated with disease with 56% of cases infected [1]. The contribution of enterotoxigenic C. perfringens to this association, was assessed by use of the reverse passive latex agglutination test for enterotoxin (RPLA; Oxoid Unipath) and vero cell toxicity neutralized by antitoxin on stored faecal samples and sporulated faecal isolates of C. perfringens. Polymerase chain reaction (PCR1) based on the DNA sequence for the whole enterotoxin gene [2] yielded a fragment from an equine isolate of the anticipated size which, cloned into plasmid M13 phage, had a sequence essentially identical to the published sequence. Consequently, all faecal isolates were also tested by PCR1 and for a part of the enterotoxin gene (PCR2). Significant association with diarrhoea (controls not in contact with cases) was found with positive RPLA tests on faeces (OR = 13, P = 0.002) and isolates (OR = 4.57, P = < 0.0001), vero cell toxicity of isolates (OR = 1.78, P = 0.026), and PCR1 (OR = nd, P = 0.029) but not PCR2 or vero cell toxicity of faeces. Significant association with diarrhoea was also found for isolates negative by RPLA (OR = 3.91; CI 2.05-7.57; P < 0.0001) or PCR1 (OR = 4.81; CI 2.84-8.20; P < 0.0001). Many of the isolates from RPLA positive faeces and verotoxic isolates were PCR negative and no evidence could be found for the presence of the enterotoxin gene in a random selection of RPLA positive/PCR negative isolates by gene probe on chromosomal DNA and PCR reaction product or vero cell toxicity neutralized by specific antiserum. Failure of the vero cell toxicity on faeces to be associated with diarrhoea or for cytotoxicity of cultures and RPLA on cultures to agree with the PCRs was believed to be related to the presence of other cytotoxins, the inherent cytotoxicity of equine faeces and to the poor specificity of the commercial antiserum used in the test. Enterotoxigenic C. perfringens could not account for the overall association of C. perfringens with foal diarrhoea because (a) cultures positive by PCR, RPLA or cytotoxicity were not significantly more common amongst isolates from cases than controls; and (b) the proportion of isolates from cases positive by PCR (PCR1 or PCR2) was too small at 9.7%.