The reactive effects of concurrent verbalization (CV) on task performance in impression formation research have not been investigated, despite increasing use of this process tracing method. Since many person perception tasks involve multiple trials, assessment of reactivity should focus on carry-over and reactive-practice effects, as well as changes in task performance concurrent with thinking aloud. An experimental method for assessing these three forms of reactivity was demonstrated in an information-request impression formation task. Concurrent reactive effects were found for both continuous and intermittent CV. For the former, task processing was mildly slowed but altered in a manner suggestive of more efficient learning of, and memory for cues requested early in task performance. The substantial slowing of task processing during trial 1 intermittent CV largely disappeared by trial 2, indicating that subjects may require more practice for this method to be used effectively. The results for continuous CV are in line with recent findings which are inconsistent with certain predictions from the predominant theory of verbal protocol generation (Ericsson and Simon, 1984). Investigators using CV are urged to incorporate empirical checks for reactivity into their experimental designs.