PURPOSE To evaluate the diagnostic performance of minimum apparent diffusion coefficient (ADCmin ) and lesion-to-spinal cord signal intensity ratio (LSR) in the differentiation of benign and malignant pulmonary lesions. MATERIALS AND METHODS Forty-seven patients (36 men, 11 women; range, 17-81 years) with 62 pulmonary lesions underwent magnetic resonance imaging (MRI) and diffusion-weighted imaging (DWI) performed using a 1.5T scanner during free-breathing using b 600 s/mm2 . The ADCmin values and LSR were calculated. A receiver operating characteristic (ROC) curve analysis was performed to detect the threshold value of the ADCmin and LSR values for discrimination between benign and malignant pulmonary lesions. RESULTS There were 42 malignant (33 primary tumors, 9 metastases) and 20 benign lesions. The mean ADCmin ± standard deviations (×10-3 mm2 /s) were 1.45 ± 0.33 for malignant tumors, and 2.4 ± 0.69 for benign lesions (P < 0.001). The mean LSR ± standard deviations for lung cancer was 1.24 ± 0.78, and for benign lesions was 0.55 ± 0.57 (P < 0.001). The area under the ROC curve for ADCmin (0.931; 95% confidence interval [CI]: 0.868-0.993) was greater than that for LSR (0.801; 95% CI: 0.675-0.926) (P = 0.029). For benign/malignant discrimination, the ROC curve showed threshold value of ADCmin to be 1.78 × 10-3 mm2 /s and that of LSR to be 0.86. Using these cutoff values, accuracy of ADCmin and LSR were 89%, 74%, respectively (P = 0.383). CONCLUSION Being a contrast-free and radiation-free technique, DWI allows discrimination of benign and malignant lung lesions. The ADCmin value performed marginally better than LSR values in distinction of benign and malignant lesions. LEVEL OF EVIDENCE 1 J. Magn. Reson. Imaging 2017;45:845-854.