OBJECTIVE The purpose of this blinded study was to test the value of primary and secondary signs, independent of each other, for diagnosis of a tear of the anterior cruciate ligament (ACL) by means of MR imaging. METHODS MR images of the knee in 74 patients who had the status of the ACL confirmed arthroscopically were blindly reviewed for status of the ACL three times: with primary signs masked, with secondary signs masked, and with no signs masked. On the basis of the observed signs at each session, the status of the ACL was predicted and the confidence in that prediction was noted. The MR imaging predictions were compared with the arthroscopic findings. The results were analyzed with receiver operating characteristic curves and stepwise discriminant analysis. RESULTS We found no difference in diagnostic performance when only primary signs were used and when both primary and secondary signs were used. When using primary signs only, our observers performed significantly better than when using secondary signs alone. However, with secondary signs alone, our observers performed better than if left to chance for predicting ACL status. Of the secondary signs, our statistical analysis found bone contusion, anterior translation of the tibia, and an uncovered posterior horn of the lateral meniscus to be the most useful for diagnosis. CONCLUSIONS Secondary signs do have value for deciding ACL status independent of primary signs. However, our observers performed much better when using primary signs instead of secondary signs. In the clinical setting, secondary signs do not help significantly in the diagnosis of ACL tears by means of MR imaging.