The Emotional Intelligence (EI) Scale devised by Schutte et al. (1998) is widely used in EI research. There have been criticisms of this scale, mainly related to its preponderance of forward-keyed items; differing results on its factor structure also exist. We investigated the effect on the scale's psychometric properties of reversing some items and adding some new items. The short form of the Bar-On EQ-i was completed by the same group of participants, comprising 500 Canadian undergraduates. The use of item reversals and additional items was found not to improve the EI scale's internal reliability. In contrast to previous studies, only three factors could be identified for the new 41-item scale: Optimism/Mood Regulation, Utilisation of Emotions and Appraisal of Emotions. Overall EI measured using the 41-Item scale and the short EQ-i were highly correlated, whilst correlations between the subscales of the two tests were theoretically interpretable. (C) 2003 Elsevier Ltd. All rights reserved.