Atypical lipomatous tumor/well-differentiated liposarcomas and dedifferentiated liposarcomas are characterized by the amplification of MDM2 and CDK4 genes. To evaluate the accuracy of fluorescence in situ hybridization (FISH) analysis in the differential diagnosis of adipose tissue tumors, we investigated MDM2-CDK4 status by FISH, real-time polymerase chain reaction (PCR) [quantitative PCR (Q-PCR)] and immunohistochemistry (IHC) in a series of 200 adipose tumors. First, we evaluated MDM2-CDK4 amplification and expression in a series of 94 well-defined adipose tissue tumors. Results showed that FISH was interpretable in 45 of 50 cases (90%), and was more specific and sensitive than Q-PCR and IHC. We then used the same techniques as complementary diagnostic tools in a series of 106 adipose and soft tissue tumors of unclear diagnosis to distinguish between (i) lipomas and atypical lipomatous tumor/well-differentiated liposarcomas, (ii) malignant undifferentiated tumors and dedifferentiated liposarcomas, and (iii) a variety of benign tumors and liposarcomas. Our results indicate that although helpful, IHC alone is often insufficient to solve diagnostic problems. FISH and Q-PCR methods gave concordant results and were equally informative in most cases. However, the proportion of noninterpretable cases was slightly higher with FISH than with Q-PCR. When tumor cells represented a minor component of the tumor tissue, such as with inflammatory tumors, FISH was more powerful than Q-PCR by allowing visualization of individual cells. In conclusion, we recommend that the evaluation of MDM2-CDK4 amplification using FISH or Q-PCR be used to supplement IHC analysis when diagnosis of adipose tissue tumors is not possible based on clinical and histologic information alone.