Recent work has shown that implicit first impressions of other people can be rapidly updated when new information about them is highly diagnostic or provides a reinterpretation of the basis of prior belief. The Affect Misattribution Procedure (AMP; Payne, Cheng, Govorun, & Stewart, 2005) is one prominent implicit measure that has been widely used in this and other work. However, the status of the AMP as a measure of unintentional responding has been a matter of debate, which necessarily also raises questions about the "implicitness" of the updated responses within recent person impression research. In re-analyses of published work, we identify multimodal distributions of AMP responses that raise concerns about potential intentional influences on this task. Drawing on 8 new studies, however, we find that such patterns are not likely attributable to intentional responding (Studies 1, 2A-2B), and that methodological modifications to the AMP procedure eliminate bimodality but do not eliminate effects of rapid revision (Studies 3A-6). Furthermore, these modifications provide evidence that the rapid-revision effects reported in earlier work can be produced under suboptimal conditions such as distraction and increased vigilance against prime influence. We advocate for the continued use of judgmental misattribution as a valuable tool in the arsenal of implicit social cognition researchers, but also encourage researchers to continue to examine the distributional patterns of measures like the AMP, and what those patterns might reflect. (PsycINFO Database Record (c) 2019 APA, all rights reserved).