Circulating tumor cells (CTCs), defined as tumor cells circulating in the peripheral blood of patients with solid tumors, are relatively rare. Diagnosis using CTCs is expected to help in the decision-making for precision cancer medicine. We have developed an automated microcavity array (MCA) system to detect CTCs based on the differences in size and deformability between tumor cells and normal blood cells. Herein, we evaluated the system using blood samples from non-small-cell lung cancer (NSCLC) and small-cell lung cancer (SCLC) patients. To evaluate the recovery of CTCs, preclinical experiments were performed by spiking NSCLC cell lines (NCI-H820, A549, NCI-H23 and NCI-H441) into peripheral whole blood samples from healthy volunteers. The recovery rates were 70% or more in all cell lines. For clinical evaluation, 6 mL of peripheral blood was collected from 50 patients with advanced lung cancer and from 10 healthy donors. Cells recovered on the filter were stained. We defined CTCs as DAPI-positive, cytokeratin-positive, and CD45-negative cells under the fluorescence microscope. The 50 lung cancer patients had a median age of 72 years (range, 48–85 years); 76% had NSCLC and 20% had SCLC, and 14% were at stage III disease whereas 86% were at stage IV. One or more CTCs were detected in 80% of the lung cancer patients (median 2.5). A comparison of the CellSearch system with our MCA system, using the samples from NSCLC patients, confirmed the superiority of our system (median CTC count, 0 versus 11 for CellSearch versus MCA; p = 0.0001, n = 17). The study results suggest that our MCA system has good clinical potential for diagnosing CTCs in lung cancer.