OBJECTIVES: Women with osteoporosis do not follow their treatment strictly and it is necessary to assess, describe and explain why. A literature review did not allow specific osteoporosis questionnaires assessing adherence (i.e. significant and self-conscious involvement of patients in their treatment) to be identified. The objective was thus to develop a tool to evaluate adherence to treatment of women with osteoporosis. METHODS: Face-to-face semi-directive interviews were conducted with 10 women with osteoporosis. Interpretative Phenomenological Analysis (IPA) was undertaken using the full transcripts of interviews. The numerous concepts identified with IPA were discussed with experts and relevant concepts for adherence were organised into a conceptual model. Items were then generated for each detailed concept using patients’ own words. The questionnaire was then tested for relevance and comprehension with 5 other women with osteoporosis, and revised accordingly. RESULTS: The conceptual model included various elements linked to adherence to treatment: personal data, beliefs, perceptions, behaviour and information received. Several items were generated for each general concept and organised into sections about osteoporosis, treatment, information received about osteoporosis and treatment, and osteodensitometry test. Except a few items that were modified or deleted, the majority of the items were well understood, and considered relevant and adequate by patients during comprehension tests. The revised questionnaire contained 44 items. CONCLUSIONS: This questionnaire is a unique and promising tool that provides a comprehensive and specific evaluation of women’s adherence to osteoporosis treatment. A future pilot study with 10 clinicians and 30 women with osteoporosis is planned in real conditions of use to assess if the questionnaire is well accepted by both clinicians and patients in clinical practice. A validation study will then be undertaken to validate psychometric properties and scoring of the questionnaire, and to demonstrate that the score is predictive of patients’ actual behaviour.