BACKGROUND: Health services research practices in the context of ethnic and cultural diversity are under scrutiny. One issue that needs to be considered is that of inclusion in, and exclusion from, research studies on the basis of language ability. We explore current reporting practice concerning recruitment of participants in health services research using examples from one 'high impact' journal and suggest ways in which future reporting can be strengthened. METHODS: Retrospective review of 207 'original research pieces' published in the BMJ. INCLUSION CRITERIA: articles published in 2003 and 2004 reporting research entailing direct communication between a researcher and health service users. A data extraction checklist was applied concerning the reporting of language-related decisions in recruitment. RESULTS: Eighty-four percent of the research articles did not engage with language issues at all. For most papers it was impossible to ascertain whether research was carried out in a monolingual population or whether researchers had effectively excluded non-primary language speakers in their recruitment procedures. Over half (n=34) of the papers that mentioned language did so in relation to exclusion criteria, usually without further comment. CONCLUSIONS: Reporting practice on language-related decisions in recruitment and on the exclusion of research participants on grounds of language ability needs to be improved. A checklist for authors, reviewers or editors is offered with the aim of encouraging fuller reporting on the language composition of source and sample populations, and greater consideration of any implications that exclusion/inclusion on the grounds of language may have for studies and their findings.