The psychological assessment of chronic pain is often accomplished using questionnaires such as the (West Haven–Yale) Multidimensional Pain Inventory ((WHY)MPI) which is constructed to capture the multidimensionality of chronic pain. The (WHY)MPI theoretically originates from behavioural and cognitive behavioural theories of pain. It is divided into three parts and measures psychosocial and behavioural consequences of pain. This questionnaire has displayed satisfactory psychometric properties and translations of the original English version into German and Dutch have been demonstrated to be reliable and valid. The aim of this study was to test the reliability and factor structure of a Swedish translation of the (WHY)MPI, the MPI-S, and also to test the generalisability of the factor structure found for the (WHY)MPI. We performed analyses of internal consistency using Cronbach's alpha, and carried out a confirmatory factor analysis (CFA) employing LISREL-8 on a population of 682 patients suffering from chronic musculoskeletal pain. Test-retest analysis was accomplished on a sub-sample of 54 individuals taken from the aforementioned population. For sections 1 and 2 of the MPI-S the overall reliability and stability were good, and after the exclusion of four items, the factor structure was similar to other versions of the MPI. For section 3, despite removal of five questions, the proposed factor structure could not be replicated. This part of the inventory is designed to measure the extent of different types of activities, and our results suggest that this section may only be used for assessing general activity level. We conclude that, with a few adjustments, the analyses yielded satisfactory results for sections 1 and 2 of the MPI-S regarding its factor structure, reliability and generalisability. For section 3 the hypothesised factor structure could not be confirmed.