Previous research has shown an inconsistent pattern of how oxytocin (OT) affects the distinction between self and others: whereas one line of studies has revealed that OT blurs the self-other distinction, other studies have not. In an attempt to solve these inconsistencies, we hypothesized that OT blurs the boundary between self and other implicitly but not explicitly. To test this assumption, we used two experimental approaches. After participants intranasally self-administered OT or placebo, they were eye-tracked while conducting a prediction task (Study 1) or they were video-recorded while conducting a distraction task with a human counterpart (Study 2). The findings confirmed the hypothesis. People usually show a distinct eye movement pattern when making self and other predictions. OT changed this pattern: the eye movement behavior during other predictions approached the eye movement behavior during selfpredictions under OT in Study 1. In Study 2, OT made the participants to mimic their confederates' mannerisms more strongly, displaying a behavioral self-other merge. Importantly, these incidents of self-other blurring only emerged implicitly. In Study 1, conscious likelihood estimates sharply differentiated between other and selfpredictions, and participants in Study 2 were not aware of mimicking the others' behaviors. Thus, a self-other merge did not occur explicitly. These findings provide some insights into recent inconsistencies of OT research and add to the understanding of the nonapeptide itself.