Ossifying fibroma (OF) and fibrous dysplasia (FD) are two fibro-osseous lesions with overlapping clinicopathological features, making diagnosis challenging. In this study, we applied a whole-genome shallow sequencing approach to facilitate differential diagnosis via precise profiling of copy number alterations (CNAs) using minute amounts of DNA extracted from morphologically correlated microdissected tissue samples. Freshly frozen tissue specimens from OF (n = 29) and FD (n = 28) patients were obtained for analysis. Lesion fibrous tissues and surrounding normal tissues were obtained by laser capture microdissection (LCM), with ~30-50 cells (5 000-10 000 µm2) per sample. We found that the rate of recurrent CNAs in OF cases was much higher (44.8%, 13 of 29) than that in FD cases (3.6%, 1 of 28). Sixty-nine percent (9 of 13) of the CNA-containing OF cases involved segmental amplifications and deletions on Chrs 7 and 12. We also identified eight CNA-associated genes (HILPDA, CALD1, C1GALT1, MICALL2, PHF14, AIMP2, MDM2, and CDK4) with amplified expression, which was consistent with the copy number changes. We further confirmed a jaw lesion with a previous uncertain diagnosis due to its ambiguous morphological features and the absence of GNAS mutation as OF based on the typical Chr 12 amplification pattern in its CNA profile. Moreover, analysis of a set of longitudinal samples collected from an individual with a cellular lesion in suspicion of OF at the first surgery, recurrence and the latest malignant transformation revealed identical CNA patterns at the three time points, suggesting that copy number profiling can be used as an important tool to identify borderline lesions or lesions with malignant potential. Overall, CNA profiling of fibro-osseous lesions can greatly improve differential diagnosis between OF and FD and help predict disease progression.