PURPOSE: This study compared the sensitivity of two commercial computer-aided detection (CAD) systems in identifying noncalcified pulmonary nodules on low-dose multidetector computed tomography (MDCT) scans by using a double reference standard. MATERIALS AND METHODS: Three chest low-dose MDCT scans of patients who had undergone lung cancer screening were retrospectively analysed using two distinct commercial CAD systems: LungCAD VC10A, Siemens Medical Solutions (CAD1) and LungVCAR, GE Healthcare (CAD2). The exact location of each finding suggested by each system was recorded by an independent reader according to spatial coordinates (x, y, z). Two panels of experienced thoracic radiologists from two different institutions independently established two reference standards (RS1, RS2) by identifying the true positive findings with spatial coordinates without using CAD. Sensitivity of the two CAD systems, defined by lesionlevel analysis, was tested and sensitivities compared. RESULTS: RS1 identified 34 noncalcified pulmonary nodules, whereas RS2 identified 54. The total number of findings detected by the two CAD systems was 684. CAD1 correctly identified 13/34 nodules (sensitivity 38%) for RS1 and 17/54 (sensitivity 30%) for RS2, whereas CAD2 correctly identified 11/34 nodules (sensitivity 35%) for RS1 and 13/54 (sensitivity 23%) for RS2. Comparison between the two CAD systems did not show a statistically significant difference in terms of sensitivity (p<0.05) for both RS1 (p=0.42) and RS2 (p=0.33). CONCLUSIONS: The two commercial CAD systems had similar sensitivity in detecting noncalcified pulmonary nodules on low-dose MDCT of the chest.