Patient questionnaires are commonly used to assess patient satisfaction. This study reports on methodological experiences based on practical use of a Swedish questionnaire. The material consists of questionnaires from five different studies at some 60 wards in three hospitals. Four of the studies were performed by "routine procedure", while one was performed specially to study sampling, non-response and other losses. The results showed that a large number of patients were not given a questionnaire despite the fact that they should have been included according to the sampling criteria. In the special study barely half of those discharged answered a questionnaire corresponding to only about one in four in some studies performed routinely. Many of the patients excluded were probably in a difficult situation and their needs ought to be particularly noticed. This was true of patients who were old or confused, had language difficulties, were seriously ill, or who died during the care episode.