Background Cold (photon-deficient) vertebrae are frequently found in Tc-99m MDP bone scans especially after tumor irradiation (RT) but little has been systemically evaluated and documented for the phenomenon seen on the F-18 FDG PET (FDG PET). To evaluate causes of this finding and its significances, the FDG PET of 230 consecutive patients have been reviewed retrospectively. : Methods: The FDG PET of 230 patients was retrospectively reviewed for the possible cold vertebrae. Locations of cold vertebrae were divided into cervical, thoracic, lumbar, and/or sacral regions. Causes of cold vertebrae were divided into 2 categories: malignant and non-malignant. Types of malignancies were also analyzed. Those with RT, FDG PET studies were performed at least 2 months’ duration after RT. As compared to the adjacent normal vertebral activity, photon-deficient areas in the vertebrae (or “cold” vertebrae) were defined by visual inspection and under the agreement of 2 nuclear medicine physicians. The FDG PET images of these patients were further correlated with their radiological X-ray, CT, and/or MR. Results: Forty-one out of the 230 (17.83%) cases showed cold vertebrae. Thirty-seven of them (90.24%) caused by effects of RT due to cancer therapy while the other 4 were non-malignant (1 hemangioma, 2 compression fractures and one false positive due to scoliosis).The malignant categories (average aged 72.2 yrs) included 20 lung malignancies (17 NSCLC and 3 SCLC); 11 head and neck squamous cell carcinomas in the supra-glottis,larynx, soft palate, tongue, vocal cord, and lip; 3 rectal carcinoma; 1 esophageal carcinoma; 1 Hodgkin lymphoma and 1 prostate carcinoma. There is evidence of overlapping effects of RT that exhibited reduced or absent bone marrow activity around the irradiated fields. The 4 non-malignancies (with non-RT; average aged 74.5 yrs) including 1 hemangioma in the T-7, 8 spines; 2 old compression fractures in the lumbar-sacral vertebrae and 1 scoliosis in the lumbar vertebrae miss-interpreted as cold vertebrae on the sagittal views. However, for detecting cold vertebrae, sagittal sections appeared offering the best view for interpretation; alternatively coronary sections can be used in certain circumstances. Conclusion: The majority (nearly 90%) of cold vertebrae is caused by RT to treat malignancies that could be early detected by FDG PET. Certain benign natures however, might also cause a similar pattern.