Subjects were asked for either free or ordered recall of eight-consonant immediate memory stimuli; ordered recall was scored by both free- and ordered-recall criteria. In terms of total errors, free-recall scoring performance was better than ordered-recall scoring; however, instructed free recall was better than ordered recall scored by a free-recall criterion. In all cases, emission of a redundant prefix increased error frequencies. There was evidence that with regard to the shape of the serial position curve, the instructional set was more important than the scoring criterion.