Between January 1, 1931, and January 31, 1970, 121 patients were treated for chondrosarcoma of the pelvis and proximal end of the femur. Of these 121 patients, 113 with 152 operations met stated criteria and were used to study factors influencing survival rate. At the time of follow-up, forty patients were alive and seventy-three were dead. Of the forty living patients, five had disease and thirty-five showed no evidence of disease—eleven of the thirty-five more than fifteen years after their last surgical procedure. Of the seventy-three dead patients, fifty-nine had died from their chondrosarcoma and fourteen, from other causes. The data analyzed were sex, age, preoperative duration of a recognized mass, preoperative duration of pain, site and size of the lesion, histological grade of the tumor, and date and type of surgical treatment. The basic method of analysis was by means of survival curves. The significance of the differences between curves was tested by the Wilcoxon-Gehan test. The findings in this study indicate that histological grading is of prognostic importance and that in the surgical treatment of chondrosarcomas in this region there must be a wide margin of normal tissue between the tumor and the line of resection if the best cure rate is to be attained.