Two issues regarding associations between negative emotions are addressed: (i) whether different types of data imply similar associations and (ii) whether these associations can be accounted for by the overlap between emotion components. In study 1, issue (i) was examined using individual differences covariation, situational covariation, and judged emotion similarity data. Results showed that the associations implied by the different data types were highly similar. Regarding issue (ii), we argue theoretically that component overlap may account for the obtained correspondence. Study 2 empirically demonstrated that the different types of association were strongly related to overlap in appraisals and action tendencies. Similarity judgments, however, seemed to be primarily based on emotion co-occurrence, and were only to a smaller degree directly related to component overlap. Copyright © 2004 John Wiley & Sons, Ltd.