Chronic antral gastritis, Lewis(a+) phenotype (Le(a+)), and male sex are common in patients with peptic ulcer. To approximate the relative risks (RR) and possible interactions of these factors in predicting coexisting active duodenal (DU) or gastric ulcer (GU), a consecutive endoscopic series of 140 ulcer patients and 215 non-ulcer controls was examined. The Lea phenotype (Le(a+) versus Le(a-)) was determined immunohistochemically as binding of Le(a)+-specific monoclonal antibody to surface epithelial secretory mucosubstances in gastric biopsy specimens. The presence versus absence of the gastritis was determined histologically from antral specimens. The RRs of the factors in the prediction of ulcer were approximated as age-adjusted RRs when the risk of ulcer in the absence of the factors--that is, in the absence of gastritis, in female sex and in Le(a-) phenotype--was applied as a base line (RR = 1). A case-control design, logistic linear modelling, and the maximal likelihood method were used in estimation of the risks. The RR of coexisting distal ulcer (DU or pyloric or prepyloric GU) was increased in the presence of gastritis (RR = 10.2), in male sex (RR = 3.0), and in Le(a+) phenotype (RR = 1.8). The RR of proximal ulcer (angular or corpus GU) was increased in the presence of gastritis (RR = 35) but decreased in the presence of male sex (RR = 0.5) and Le(a+) phenotype (RR = 0.7). As predictors of both distal and proximal ulcer, gastritis, sex, and Le(a) phenotype were independent of each other; that is, their joint value in prediction of ulcer is a multiplicand of the marginal risks. Thus, a 50-fold difference in the joint RR could be approximated between the extreme risk groups for distal ulcer--that is, between Le(a+) males with gastritis and Le(-a) females with normal antrum. In a consecutive series of outpatient endoscopies, 45% of females and 8% of males could be categorized to these extreme 'low'- and 'high'-risk groups, respectively. We conclude that sex, Le(a) phenotype, and gastritis are factors that, at least in ordinary outpatient endoscopy material, divide subjects to subgroups with very different risks and probabilities for having coexisting peptic ulcer.