Suspected pelvic endometriosis was prospectively evaluated in 31 women with T1- and T2-weighted conventional spin-echo (CSE) magnetic resonance imaging alone and in combination with T1-weighted fat-suppressed (T1FS) and gadolinium-enhanced T1FS (Gd-T1FS) spin-echo techniques. Images were grouped for interpretation and comparison as follows: (a) CSE alone, (b) CSE/T1FS, and (c) CSE/T1FS/Gd-T1FS. All patients underwent surgery within 3 months of imaging, and 21 patients were found to have endometriosis: 59 endometriomas (26 large and 33 small) and 51 sites of implants were seen. With CSE images, 23 large and six small endometriomas were detected. With CSE/T1FS images, 25 large and 14 small endometriomas were identified. With CSE/T1FS/Gd-T1FS images, 24 large and 14 small endometriomas were detected and ill-defined areas of enhancement were noted in 22 sites throughout the pelvis. These corresponded to endometriotic implants seen at surgery in 14 sites. The sensitivity, specificity, and accuracy, respectively, for the detection of endometriosis were 76%, 60%, and 71% for CSE, 86%, 50%, and 74% for CSE/T1FS, and 81%, 50%, and 71% for CSE/T1FS/Gd-T1FS images. No significant differences (P > .1) between image combinations for correctly identifying patients with and without endometriosis were seen. The difference in sensitivity between CSE and CSE/T1FS and between CSE and CSE/T1FS/Gd-T1FS images for detecting small endometriomas was significant (P = .03).