The tumor necrosis factor alpha (TNFA)-308*A allele has been found to confer an increased risk of gastric carcinoma. Inconsistency in risk estimates across populations lead us to hypothesize about the presence of an alternative causal locus in the same chromosomal region. A suitable approach is to determine the tumor necrosis factor haplotypic structure in order to clarify whether the association between the *A allele and the increased risk of gastric carcinoma is etiologic or secondary to linkage disequilibrium. Firstly, we assessed the association between the TNFA-308G>A polymorphism and the risk of gastric carcinoma in a population from Northern Portugal (508 gastric carcinoma patients, 713 controls); secondly, we genotyped five microsatellite loci (TNFa, b, c, d, e) flanking the TNFA-308G>A locus to establish the haplotypic structure associated with this single-nucleotide polymorphism in cases (122 patients) and controls (169 individuals). We found a significant association between the *A allele and increased risk of gastric carcinoma (odds ratio, 1.7; 95% confidence interval, 1.3-2.2) confirming previous results in our population. Regarding the *A allele-associated haplotypes, the most relevant difference was found for the H1A haplotype present in 33.1% of the cases and 12.5% of the controls. We also observed haplotypes associated with the *A allele that were found only in cases or controls. A population differentiation test showed that the gastric carcinoma and the control groups were significantly different for the *A allele haplotypic structure. This suggests that the association between the TNFA-308G>A polymorphism and increased risk of gastric carcinoma is dependent on linkage disequilibrium with an as yet unidentified locus.