This paper discusses a development of Flanagan's critical incident technique (CIT) to elicit indicators of high and low quality nursing from patients and their nurses on medical, surgical and elderly care wards. Stages in undertaking the CIT are identified and presuppositions held by most researchers about the nature of the technique are identified. The paper describes how the authors moved to a different set of presuppositions during the course of the study. Preliminary analysis of interview transcripts revealed that critical incidents need not always be demarcated scenes with a clear beginning and end, but may arise from respondents summarizing their overall experience within their description of one incident. Characteristically respondents were unable to give a detailed account of such incidents but validity may be established by the fact that respondents appear to recount what actually happened as they saw it, and what they said was clearly important to them. The researchers found that the most appropriate basic unit of analysis was not the incident itself but 'happenings' revealed by incidents that are 'critical' by virtue of being important to respondents with respect to the quality of nursing care. The importance of CIT researchers achieving an understanding of the 'meaning' of critical happenings to respondents is emphasized. Analysis of the interview transcripts is facilitated by the use of INGRES, a relational database computer program which should enable a 'personal theory' of quality nursing for each respondent, both patients and nurses, to be described. The study suggests that the CIT is a flexible technique which may be adapted to meet the demands of nursing research. If carefully applied, the CIT seems capable of capitalizing on respondents' own stories and avoids the loss of information which occurs when complex narratives are reduced to simple descriptive categories. Patients and nurses have unique perspectives on nursing and their views are of primary importance in understanding what quality means with respect to the interpersonal processes that are integral to nursing care. This paper discusses the identification of indicators of quality nursing from interviews with patients and nurses using the authors' development of Flanagan's critical incident technique.