Very few recent attribution papers have investigated the effect of outcome on athletes attributions. The present study intended to examine any differences in these attributions during both successful and unsuccessful performances. Previous research (e.g. Gernigon & Delloye, 2003; Stoeber & Becker, 2008) found performance outcome to influence attributions and in turn, other concepts such as self-efficacy. Consequently, it was hypothesised that there would be an interaction between attribution used and performance outcome. Most, if not all, research has used questionnaires to measure attributions post event, encountering potential problems of retrospective recall (Stone et al. 1998) and memory decay (Smith, Leffingwell & Ptacek, 1999). A potential way to counteract this may be to use Think Aloud (TA) protocol to capture attributions in-event. The study will be the first to use TA to capture in-event attributions. Nine competitive golfers (handicap M = 6, SD = 3.64) from Hampshire were trained in TA and given a 15 shot short game task, including chips, pitches and putts of varying difficulty. Attributions were recorded, transcribed verbatim and coded using Weiner (1972) attributions, and determined functional or not according to Rudisill (1989). Number of successful or unsuccessful shots and number of attributions produced were analysed preliminarily. Following this, functionality and each of the attributional dimensions were analysed using 2 x 2 repeated measures analysis of variance (ANOVA). Preliminary analyses revealed a significant difference in number of successful and unsuccessful shots, but no significant difference in attributions produced by successful and unsuccessful shots. Main analyses revealed no significant interaction at all. However, significant main effects were found for outcome when analysing functionality, as well as for the three attributional dimensions analysed. The hypothesis of an interaction between performance outcome and attribution used was rejected. This is inconsistent with previous literature, but may be due to factors such as a small sample size (Cohen, 1992) or the fact it is the first measurement of in-event attributions. Differences in the attributional dimensions verbalised is consistent with previous research (e.g. Gernigon & Delloye, 2003). Future research should continue to use TA to measure in-event attributions, across different sports, genders and abilities. Researchers should look to develop a measurement tool for in-event attributions, to enable easier comparison across studies.