At many institutions, three-dimensional magnetic resonance imaging (MRI) is routinely used for examination of the knee. Multiplanar reconstruction (MPR) is a method of displaying three-dimensional datasets. The authors assessed the usefulness of MPR for evaluating knee MRI datasets by comparing readers' performance using MPR and conventional film MRIs. Eight patients with internal derangement of the knee were studied. All had MRI datasets acquired in the sagittal plane using a three-dimensional gradient-echo fast imaging with steady-state precession (FISP) sequence. Arthroscopic surgery after MRI confirmed the presence of 6 anterior cruciate ligament (ACL) tears, 11 meniscal tears, and 5 normal menisci in this group. Four blinded readers, experienced in MRI of the knee, evaluated the images. The MRI datasets were then loaded onto a three-dimensional workstation and interpreted by the same readers using MPR. The MRI findings were correlated with arthroscopy. For diagnosis of tears of the ACL, sensitivity was 96% and specificity was 100% for both films and MPR. For detecting meniscal tears, sensitivity was 55% and specificity was 90%, using the filmed images, versus 64% and 85%, respectively, with MPR. These differences were not statistically significant by the sign test. Total time (technologist processing time plus radiologist reading time) for MPR was greater than for film interpretation by a factor of 1.12 (P < .05), and radiologist reading time for MPR was greater by a factor of 1.88 (P < .05). For sagittal three-dimensional FISP MRI datasets, real-time MPR is comparable with film interpretation for evaluation of ACL and meniscal injuries, but it increases the time required for diagnosis.