The recent successes being achieved with combination chemotherapy regimens strongly indicate that gastric cancer is chemosensitive. With these regimens, objective remission rates of more than 50% were recorded, including approximately 10% complete remission (CRs). The finding that locally advanced disease (LAD) and technically unresectable disease could be rendered resectable by preoperative chemotherapy (EAP) was important. The median survival time was 18 months for all patients and 24 months for disease-free patients. At 30+ months, 21% of all patients are still living disease free. The expected survival of patients with unresectable LAD is approximately 4-6 months without any treatment and 6-9 months with standard chemotherapy. Compared with the latter results, the preoperative use of etoposide, adriamycin, and platinum (EAP) seems to improve the prognosis of patients with LAD. Moreover, such a multimodal approach may increase the number of long-term survivors among patients with resectable gastric cancer, especially those whose stage indicates a high risk of relapse (stage IIIa or IIIb). However, partly because of the severe toxicities (myelo-suppression, nausea/vomiting), a considerable number of patients cannot be treated with these new regimens. Considering the predominantly palliative treatment intentions in far-advanced (metastasized) gastric cancer, regimens with low toxicities and acceptable activity should be preferred. For these reasons, we developed and investigated the combination etoposide, leucovorin, and 5-FU (ELF) in a phase II trial in elderly patients (greater than 65 years) and in patients with cardiac risks who could not be treated with anthracyclines. The overall response rate in 51 evaluable patients was 53%, including six clinical CRs (12%). The median remission duration was 9.5 months and the median survival time was 11 months. Tolerability was excellent. The high remission rate and long medical survival time achieved with ELF, plus its good tolerability, make this combination a valuable alternative to anthracycline-containing regimens. In addition to developing new treatment plans and strategies, it is also desirable to predict treatment outcome with an acceptable degree of certainty. Combinations of variable defined patient subgroups with significantly different probabilities for achieving objective response and for survival. The results of this analysis may contribute to a more patient-oriented and risk-adapted chemotherapy and to better comparability of future studies.