The relative utility of various preoperative diagnostic imaging modalities, including PET (utilizing FDG and FMT), CT, and MR imaging, for evaluation of lipoma and liposarcoma, especially well-differentiated liposarcoma, was investigated. Imaging findings in 32 patients with histopathologically documented lipoma, including one with fibrolipoma and one with angiolipoma, and 25 patients with liposarcomas whose subtypes included 10 well-differentiated, 10 myxoid, and 5 other types were reviewed retrospectively. Pre-operative imaging included FDG-PET (n = 44), FMT-PET (n = 21), CT (n = 25), and MR imaging (n = 53). Statistically significant imaging features of MR images favoring a diagnosis of liposarcoma involved lesions containing less than 75% fat (p < 0.001) as well as the presence of septa (p < 0.001). As compared with well-differentiated liposarcoma, benign lesions were differentiated significantly only by the presence of septa (p < 0.001), which also provided significant differentiation on CT (p < 0.05). The mean SUVs for malignant tumors were significantly higher than those for benign lesions in both FDG- and FMT-PET analyses (p < 0.0001, p = 0.0011, respectively). By using a cut-off value for FDG- and FMT-PET set at 0.81 and 1.0 respectively, which provided the highest accuracy, benign lesions were differentiated significantly from liposarcomas (p < 0.001, and p < 0.02). Furthermore, benign tumors and the three subtypes of liposarcoma were divided significantly into four biological grades by FDG- and FMT-accumulation rates (rho = 0.793, p < 0.0001; and rho = 0.745, p = 0.0009, respectively). A cut-off value of 0.81 for FDG-PET provided significant differentiation between benign lesions and well-differentiated liposarcoma (p < 0.01). The presence of septa on MR images differentiated lipomas from liposarcoma, even well-differentiated type. PET analysis, especially FDG-PET, quantitatively provided not only the differentiation but also the metabolic separation among subtypes of liposarcoma. Interpretation of the visual diagnostic modalities requires extensive experience and carries a risk of ignoring a critical portion of malignancy. PET metabolic imaging may be an objective and useful modality for evaluating adipose tissue tumors preoperatively.