Research exploring spontaneously generated self-talk has involved recording performers’ self-talk categorized by researchers. The actor-observer bias, suggests that actors (performers) and observers (researchers) may perceive the same situation (e.g., self-talk) differently. The purpose of this study was to explore the actor-observer bias and validity of self-talk categorization. College students’ (n = 30) spontaneous self-talk was audio recorded during a dart throwing task. Participants then listened to and categorized their self-talk. Three independent researchers reviewed written transcripts and categorized the self-talk. Another three researchers who had not read the transcripts listened to audio recordings and categorized the same self-talk. Results confirmed actor-observer bias predictions. Spontaneous self-talk ratings made by participants were similar to but distinct from those made by researchers reading transcripts or listening to self-talk audio recordings. These results suggest that participant categorization of spontaneous self-talk may be a valid strategy to enhance understanding of self-talk used in competitive settings.