Self-efficacy - an individual's judgement of their ability to successfully perform a behaviour - is commonly used to explain and predict behaviour. It is measured through self-report questionnaires. These scales require good content validity, that is must measure the full scope and content of the construct without contamination from similar constructs. This study uses a systematic, transparent quantitative method (discriminant content validation, DCV) to assess the content validity of a variety of self-efficacy items and qualitatively explores participant interpretations of these items. A quantitative DCV and qualitative think-aloud study of self-efficacy item interpretation. Participants (n = 21) were presented with items designed to measure self-efficacy and related constructs following standard DCV methodology. Items were rated against construct definitions to determine whether they measured a particular construct (yes/no). Judges' confidence in each assessment was also assessed (%) and used to establish quantitative estimates of content validity for each item. A qualitative think-aloud study explored the judgements made in a subset of participants. 8/8 self-efficacy items were found to measure self-efficacy; however, 2/8 of these also measured motivation. 6/8 items displayed discriminant content validity and thus can be considered 'pure' measures of self-efficacy. The think-aloud study indicated that item wording is a likely cause of item misinterpretation. Self-efficacy items vary in terms of their content validity with only some of the items assessed providing 'pure' measures of the self-efficacy construct. Item wording should be considered during study design to avoid misinterpretation. Statement of contribution What is already known on this subject? For decades, questions have been raised relating to the content validity of self-efficacy scales, with suggestions of possible construct contamination. Previous studies have shown that manipulation of the wording of self-efficacy scale items results in varied participant interpretations. While self-efficacy items have been found to be distinguishable from other similar constructs, it is equally important to ensure that they are uncontaminated by different constructs within the same theory. Otherwise, when the theory is used to investigate behaviour, variance attributable to self-efficacy may be attributed to a different construct or vice versa. The present study uses discriminant content validation to test this and a think-aloud study to explore participants' interpretation of classic self-efficacy items. What does this study add? The study uses discriminant content validity methodology to assess the content validity of self-efficacy scales. Self-efficacy items which are contaminated with content from other constructs are identified. Information is presented on 'pure' self-efficacy items which can be used to guide item selection in future studies.