A new approach to distinguish normal fat, well-differentiated (WD), and dedifferentiated liposarcoma (LS) tumors is demonstrated, based on the use of optical coherence tomography (OCT). OCT images show the same structures seen with conventional histological methods. Our visual grading analysis is supported by numerical analysis of observed structures for normal fat and WDLS samples. Further development could apply the real-time and high resolution advantages of OCT for use in liposarcoma diagnosis and clinical procedures.