In an attempt to define the role of initial surgical resection in patients with undifferentiated small cell carcinoma of the lung, we reviewed the experience of the Veterans Administration Surgical Oncology Group (VASOG). One hundred forty-eight patients with small cell carcinoma of the lung had undergone a potentially "curative" resection. This represented 4.7% of "curative" resections carried out in four major prospective adjuvant chemotherapy trials. In the early trials (101 patients), 16 patients (15.8%) died within the first 30 postoperative days. These patients have been excluded from the analysis of long-term survival, since in the more recent trials postoperative deaths were excluded prior to randomization. In the 132 patients remaining, the 5 year survival rate by the life-table method was 23.0%. The tumor of each was classified pathologically by the TNM system. Five-year survival rates for each category were as follows: T1 N0 M0 59.9%, T1 N1 M0 31.3%, T2 N0 M0 27.9%, T2 N1 M0 9.0%, and any T3 or N2 3.6%. The effect of postoperative adjuvant chemotherapy was evaluated in each of the trials. No beneficial effect of the adjuvant therapy was noted with a one or two course regimen of either nitrogen mustard or cyclophosphamide, but possible benefit, although not significant, was noted in a prolonged intermittent chemotherapy trial of cyclophosphamide either alone or alternating with methotrexate. In the most recent trial of prolonged intermittent courses of 1-(2-chlorethyl)-3-cyclohexyl-l-nitrosourea (CCNU) and hydroxyurea, a 5 year survival rate of 80.8% was noted in those receiving adjuvant chemotherapy as compared to a 38.1% in the control group. We conclude that resection is definitely indicated in patients with T1 N0 M0 lesions and probably indicated in those with T1 N1 M0 or T2 N0 M0 lesions. Primary surgical resection is contraindicated in patients with any other TNM category.