Gastric carcinoma continues to claim an increasing number of victims (approximately 900 000 per year) partly due to the ageing of the population (Munoz and Franceschi, 1997), but as we enter the 21st century, the aetiology of the world's second most common cancer (Parkin et al, 1993) remains relatively poorly understood. Risk factors for the different subtypes, both according to tumour histology (intestinal/diffuse) (Lauren, 1965) and site (cardia/noncardia), need to be identified.The remarkable global decline in gastric carcinoma incidence has mainly been attributed to intestinal type (Correa and Shiao, 1994;Lauren and Nevalainen, 1993;Munoz and Asvall, 1971) although firm consensus is lacking. Its variation over time and across populations (Coleman et al, 1993) is presumed to have a predominantly environmental aetiology. In contrast, the diffuse subtype is usually postulated to be more genetically predetermined (Correa and Shiao, 1994). Reports of a disparate incidence trend for cardia tumours (Blot et al, 1991;Powell and McConkey, 1992) strongly suggest that this subsite has a distinct aetiology.The concurrently falling prevalence of infection with Helicobacter pylori (H. pylori) (Haruma et al, 1997;Parsonnet et al, 1992;Roosendaal et al, 1997;Xia and Talley, 1997), suggests that its association with non-cardia gastric carcinoma is confined to the intestinal subtype. This hypothesis is in line with Correa's model of H. pylori-induced carcinogenesis (Correa, 1995). A stable incidence of the diffuse subtype would contradict an important role of H. pylori.We conducted a prospective population-based study over 6 years, with an extensive search for all incident gastric cancer cases in well-defined populations and a strictly uniform classification of tumour site and histologic type. Our aim was to analyse the incidence of the intestinal and diffuse types separately, while taking subsite into account. Sweden, with a two-fold variation in gastric cancer risk within the country and declining gastric cancer rates, is a particularly suitable environment for studying these questions. MATERIALS AND METHODS SubjectsAll new cases of gastric adenocarcinoma diagnosed before death, from 1989 through 1994, occurring in two northern high-risk counties (Norrbotten, Västerbotten) and three southern lowaverage-risk counties (Södermanland, Uppsala, Västmanland), with a total population 1.3 million, were identified through: contacting clinicians at all hospitals; surveillance of all suspected gastric cancer specimens at the departments of pathology; monthly double-checks with all regional cancer registers; and record linkages with the nation-wide Swedish cancer and death registries. All Decreasing incidence of both major histologic subtypes of gastric adenocarcinoma Ð a population-based study in Sweden Summary While the overall incidence of gastric cancer has fallen, presumably to a large extent in parallel with Helicobacter pylori infection, the occurrence of the diffuse histologic type is thought to have remained more stable, quest...