To identify factors associated with adherence to medication for rheumatoid arthritis or undifferentiated inflammatory arthritis using a systematic literature search. PubMed, PsycINFO, EMbase and CINAHL databases were systematically searched from inception to February 2011. Articles were included if they addressed medication adherence, used a reproducible definition, determinants and its statistical relationship. Methodological quality was assessed using a quality assessment list for observational studies derived from recommendations from Sanderson et al. (2007) . Resulting factors were interpreted using the Health Belief Model (HBM). 18 out of 1479 identified studies fulfilled the inclusion criteria. 64 factors were identified and grouped according to the HBM into demographic and psychosocial characteristics, cues to action and perceived benefits versus perceived barriers. The belief that the medication is necessary and DMARD use prior to the use of anti-TNF had strong evidence for a positive association with adherence. There is limited evidence for positive associations between adherence and race other than White, general cognition, satisfactory contact with the healthcare provider and the provision of adequate information from the healthcare provider. There is limited evidence for negative associations between adherence and having HMO insurance, weekly costs of TNF-I, having a busy lifestyle, receiving contradictory information or delivery of information in an insensitive manner by the rheumatologist. 18 factors were unrelated to adherence. The strongest relation with adherence is found to be prior use of DMARDs before using anti-TNF and beliefs about the necessity of the medication. Because the last one is modifiable, this provides hope to improve adherence. Copyright © 2013 Elsevier Inc. All rights reserved.