The treatment of early-stage gastric lymphoma is controversial. This retrospective analysis reports on the outcome of 24 patients treated in our institution during the past 25 years. Fourteen patients had stage IEA, one patient had IEB, six patients IIEA1, and three patients had stage IIEA2 non-Hodgkin's lymphoma (NHL). Diffuse large cell intermediate-grade NHL was diagnosed in 17 patients, diffuse small cleaved cell in three patients, and diffuse mixed large and small cell lymphosarcoma, low-grade B-cell lymphoma, and unclassified lymphoma in one patient each. Fourteen patients underwent surgery, 21 had radiation therapy (XRT), and 10 patients received chemotherapy. Surgery + XRT were given to 7 patients, surgery + XRT + chemo and XRT alone were delivered to five patients each, and XRT + chemotherapy were employed in four patients. Surgery alone was the initial treatment in two patients and chemotherapy alone was given to one patient. Following treatment 22/24 achieved a complete response. During a mean follow-up period of 77.6 months (range 1-285), five patients relapsed. At 10 years, the actuarial survival of the 15 patients with stage I disease was 57.4% and for stage II it was 51.9% (Gehan P-value 0.33). Freedom from relapse (FFR) was 60.7% and 58.3%, respectively (P-value 0.56). No significant statistical differences in terms of survival and FFR were noted in patients treated with surgery, chemotherapy, or XRT. The outcome of patients treated with triple-modality therapy was similar to those treated with double-modality therapy and to patients treated with XRT alone. Gender, age, presenting symptoms, depth of tumor through the gastric wall, and stage were not statistically significant for prediction of either survival or FFR. Both surgery + XRT and chemotherapy + XRT are effective in the treatment of early-stage gastric disease. XRT alone is equally effective as two or three modality treatments in the subset of patients with early-stage gastric lymphoma. However, the low number of patients treated with various approaches over a long period precludes a firm conclusion. Until prospective randomized studies are initiated, management programs should be individually tailored.