The efficacy of "weighted" films in diagnosing grade 3 acromioclavicular (AC) sprains not evident on plain (unweighted) films was evaluated. Eighty-three pairs of radiographs, taken with and without weights, of patients with suspected AC injury were studied. The films were read in a randomized and blinded manner by a staff radiologist. Criteria for classifying AC injuries were: grade 1, less than 3 mm (or less than 50%) difference between the AC widths with a normal coracoclavicular (CC) distance; grade 2, greater than or equal to 3 mm (or greater than or equal to 50%) difference between the AC widths with a normal CC distance; and grade 3, greater than or equal to 5 mm (or greater than or equal to 50%) difference in CC distance. In only three cases (4%) did weights cause the injured CC distance to increase and thereby unmask a grade 3 injury not evident on plain films. Further evidence that weights may not reliably elucidate the degree of AC joint injury is suggested by the fact that in several cases the weights actually caused the injured and uninjured CC distance to decrease. We conclude that the use of weighted radiographs lacks efficacy in unmasking grade 3 AC sprains on radiograph and we recommend that routine use of this technique be abandoned.