The present investigation examined effects of repeated observation of actors' achievement behavior on VTR. Several studies have shown that an empathic observer (EO) who was instructed to observe Sp's behavior from Sp's point of view tended to attribute Sp's behavior more to external factors - Luck and Task Difficulty, etc. - than an observer who observed the behavior from his own point of view (SO). We assumed a reason why such attributional discrepancy occures lies in difference between the most salient features for SO and that for EO. We derived, then, the following hypothesis in advance to the experiment: In spite of the difference of their point of view, observers would tend to ascribe the Sps' behavior more to internal factors after the second than the first observation. An experiment was conducted according to 2 x 2 x 2 factorial design: the first variable was point of view of the observation (empathic vs. standard), then second variable was sequence of observation (first vs. second), and the third variable was performance of Sps (success vs. failure). The first was a between-Ss, while the second and the third were within-Ss factors. Ss were told that the first observation was a practice trial, while the second for data. Immediately after each of the observations, they were made to make causal attribution of the Sp's performance. Maijor findings were as follow: 1. The main effect of point of view was significant: SO tendent to make more internal attribution than EO. 2. The main effect of performance was significant, too : Ss attributed Sp's success more to internal factors than failure. 3. The main effect of the sequence was significant, as well: Ss made stronger internal attribution after second than after first observation. Thus, it might be concluded that the hypothesis has been supported.