Quantitative measurement of minimal residual disease predicting recurrence in individual cancer patients is available only in very few indications, such as acute lymphoblastic leukemia, but is still missing in most solid tumors, including non-small cell lung cancer (NSCLC). MAGE-A expression levels in blood and bone marrow determined as calibrator-normalized relative ratios by quantitative multimarker real-time RT-PCR for transcript amplification of -A1, -A2, -A3/6, -A4, -A10, and -A12 in 94 patients with completely resected NSCLC were correlated with survival in a clinical study. Patients with MAGE-A expression levels ≥0.2 in at least one sample of bone marrow or blood at tumor surgery had a significantly reduced overall ( = 0.007), cancer-free ( = 0.002), and distant metastasis-free survival ( < 0.001) versus patients below 0.2 in all samples without significant difference in locoregional recurrence-free survival. The corresponding HRs (≥0.2 vs. <0.2) for death, cancer-related death, and development of distant metastasis were 2.56 [95% confidence interval (CI), 1.42-4.63], 3.32 (95% CI, 1.66-6.61), and 4.03 (95% CI, 1.77-9.18), respectively. Five-year Kaplan-Meier estimates of distant metastasis-free survival were 43% (MAGE-A ≥ 0.2) versus 87% (MAGE-A < 0.2). MAGE-A expression in blood or bone marrow at tumor surgery is an independent predictor of survival in resected NSCLC. The reliable prediction of distant metastasis in individual patients with a statistically proven impact on overall survival may help to refine patient selection for adjuvant therapy urgently needed, especially in the clinical management of elderly patients. .