The importance of compassion is widely recognized and it is receiving increasing research attention. Yet, there is lack of consensus on definition and a paucity of psychometrically robust measures of this construct. Without an agreed definition and adequate measures, we cannot study compassion, measure compassion or evaluate whether interventions designed to enhance compassion are effective. In response, this paper proposes a definition of compassion and offers a systematic review of self- and observer-rated measures. Following consolidation of existing definitions, we propose that compassion consists of five elements: recognizing suffering, understanding the universality of human suffering, feeling for the person suffering, tolerating uncomfortable feelings, and motivation to act/acting to alleviate suffering. Three databases were searched (Web of Science, PsycInfo, and Medline) and nine measures included and rated for quality. Quality ratings ranged from 2 to 7 out of 14 with low ratings due to poor internal consistency for subscales, insufficient evidence for factor structure and/or failure to examine floor/ceiling effects, test–retest reliability, and discriminant validity. We call our five-element definition, and if supported, the development of a measure of compassion based on this operational definition, and which demonstrates adequate psychometric properties.