This study's purpose was to evaluate the reliability of knee injection sites with low-volume (2-3 cc) injection typical of viscosupplementation. This was a prospective study of 131 knees randomly assigned to 3 injection groups before to knee arthroscopy: SM-superomedial, superolateral, and lateral joint line (LJL). Needles were inserted to the hub, and aspiration was attempted before injection with 3 cc diluted methylene blue (MB). Injections then were graded at the time of arthroscopy as good, fair, or poor, depending on the intra-articular staining of MB. We conclude that a lateral joint line injection site may not be reliable for routine injections of low volumes into knees, because it results in good intra- articular delivery less than half of the time, with a high incidence of soft-tissue infiltration.