The Min mouse is a model for human familial adenomatous polyposis (FAP), an autosomal dominant disease characterised by multiple adenomatous gastrointestinal polyps. The severity of the Min phenotype is modified by a locus (Mom1) on mouse chromosome 4, at a position syntenic with human chromosome 1p35-p36. The secretory phospholipase A2 (Pla2s) gene is a candidate for this modifier locus and there is evidence that a locus on human chromosome 1p35-p36 acts to modify the severity of human duodenal FAP. We have analysed the human secretory phospholipase A2 locus (PLA2G2A) for variants that could directly influence the FAP phenotype. We found no PLA2G2A variants predicted to result in functional variation in the phospholipase A2 protein. Two PLA2G2A polymorphisms were, however, discovered, one a 'silent' base change in exon 3 and another in a noncoding region. Three other variants (possible mutations) were found in non-coding regions. In 70 FAP patients from 20 families, no associations were found between the severity of duodenal polyposis and any PLA2G2A variant. One allele at the exon 3 polymorphic site did, however, occur more often then expected in patients with relatively severe colonic FAP. Although of borderline statistical significance, this association, if genuine, is likely to result from linkage disequilibrium between the PLA2G2A alleles studied and undetected genetic variation at a closely linked locus. The frequency of the alleles at both polymorphic sites has also been determined in the germ line of patients with sporadic colorectal adenomas and carcinomas and in random controls, but no differences were found among these groups. Our results suggest that PLA2G2A variants do not influence inherited or sporadic colonic tumours. A linked locus may be a modifier of human FAP, but does not influence the risk of colorectal tumours in the general population.