In a series of experiments, we have demonstrated that providing erroneous knowledge of results (KR) biased performance in the direction of the erroneous KR statement, even when that statement was clearly in conflict with available sensory information. However, when the direction of the erroneous KR was mixed within a set of trials, the bias was eliminated. The effects of consistent versus mixed erroneous KR have been obtained only across experiments. The purpose of the present experiment was to examine directly the effects of consistent versus mixed erroneous KR statements on performance bias within a single experiment. Thirty-two volunteer participants were equally and randomly assigned to one of four groups; correct KR, erroneous KR +200, erroneous KR -200, and mixed-erroneous KR. The +200 ms and -200 ms indicated the direction (+/-) and magnitude (200 ms) of the erroneous KR statement from the participant's actual error. The " mixed " group was given an equal distribution of +200 ms and -200 ms (from the actual error) erroneous KR statements across a set of trials. All participants were given 31 acquisition trials with KR on a coincidence-anticipation timing task, followed by 31 immediate retention trials, and 31 delayed (24 hours later) retention trials without KR. As with previous experiments, AE and CE data were calculated to reflect response accuracy and bias. AE and CE data were analyzed using 4 (Group) x 5 (Trial Blocks) ANOVAs for acquisition, immediate, and delayed retention. Results of AE and CE data indicated that the groups receiving correct KR and the mixed-erroneous KR performed with less error than the groups receiving consistent erroneous KR. These results indicate that erroneous KR influences performance bias when the direction of the erroneous KR is consistent but not when the direction of the erroneous KR is mixed within a set of trials. Thus, erroneous KR in and of itself does not negatively affect performance. It is only when the erroneous KR statement is consistently biased in a certain direction that performance is biased. Keyword(s): performance, research