Patients with brain metastasis after resection of non-small-cell lung cancer usually have poor prognosis. A few such patients, however, survive for long periods after surgical resection of brain metastases. To evaluate the prognostic factors in resection of solitary brain metastasis from non-small-cell lung cancer, we reviewed 24 cases undergoing resection of solitary brain metastasis after resection of the primary site from 1977 to 1993. The patient population consisted of 20 men and four women ranging in age from 40 to 75 years old (average, 57.8 years old). None of the patients had systemic metastasis except in the brain at the time of brain surgery. The overall survival rates were 12.5% at 3 years and 8.3% at 5 years after brain surgery. The longest survival periods were 11.5 years after brain surgery and 15.4 years after lung surgery. The interval between lung and brain surgery (< or =360 days vs. >360 days), differentiation of primary cancer (poor vs. moderate), size of primary site (< or =5.0 cm vs. >5.0 cm), and operation of primary site (lobectomy vs. pneumonectomy) significantly affected survival as shown by univariate analysis (P<0.05). Other clinical factors (age, gender, histology, T- and N-status, 'resectability with curative intent' of the primary site, location of the brain metastasis and postoperative radiation therapy) did not affect survival. Multivariate analysis using Cox's proportional hazards model indicated that an interval of more than 360 days between the two surgical procedures (hazard ratio = 0.2351, P = 0.0136) and lobectomy (hazard ratio = 0.5274, P = 0.0416) were independent prognostic factors. In conclusion, patients with solitary brain metastasis from non-small-cell lung cancer without other organ metastasis, in whom relapse in the brain occurred more than 1 year after resection of the primary site and in whom lobectomy was performed, should be treated surgically to maximize the chance of prolonged survival.