This is an ethnomethodological study of qualitative research interviews concerning the death of a spouse. The focus is on the accounting practices of interview participants. Methods of analysis described by Sacks, membership categorisation analysis (MCDA) and conversation analysis (CA), have been applied to the data. The analysis also draws on Sacks's discussions of storytelling. Three different but related issues are examined in the data: criticism of health professionals, assessment work and doing interview talk. MCDA has been used to document how criticisms of health professionals are produced and to examine how assessment work is done. Criticism involves setting up lay and professional identities, and recipient-design. Interviewees venture their criticisms cautiously, setting up their accounts in such a way that the hearer is co-implicated. A feature of the detailed assessment work undertaken in the accounts is the setting up of entitlements to certain experiences by interviewees, such as being with a spouse when they die. The way in which the identities of the speakers in the (interview) talk are established in the opening turns has been examined using CA. The opening request by the interviewer, 'could you tell me the story of what happened' is produced as an openended question but the response provided is skilfully tied to a story that the interviewer expects to hear. Implications of the analysis are drawn regarding the status of interview data. The value of attending to the accounting practices of participants in producing interview data is also discussed in relation to lay assessments of health care.