Taking Fibromyalgia syndrome (FMS) as an example, the article illustrates a problem that to our knowledge has not been addressed in rehabilitation research so far: According to our large dataset, a sizeable proportion of patients had to be sent home with extremely severe burdens (<2 percentile in the normal population) at discharge - in spite of good improvements during their stay. Since 2009, patients in the RehaKlinikum Bad Säckingen, an in-patient rehab center for orthopedic-rheumatic diseases, answer the questionnaire "Indicators of Rehabilitation Status" (IRES) at the beginning and the end of their stay. We analysed IRES-data of 1 803 patients with FMS (94% women). In addition to analyses of change, we determined the degrees of severity at admission and discharge on the basis of a comparison with the normative sample of the IRES. In order to predict membership of the high-risk group of patients with still "extremely severe" values at discharge, we performed binary logistic regression analyses. At admission, about 90% of the patients showed either "extreme" (65%<2 percentile) or "severe" (27% 2-10 percentile) values on the IRES summary score as well as on the scores for "psychic status", "pain", "symptoms of orthopedic and cardiovascular diseases", and "functioning in everyday life". In sum, then, FMS-patients have come to rehabilitation with multiple burdens of a severe to extreme degree. At discharge, the mean summary score had improved with a "strong" effect size of SRM=1.07. In spite of these good overall improvements, however, 37.4% of the patients went home with "extreme" burdens remaining, even though almost 60% of them had experienced "strong" (28%) or "relevant" (31%) improvements. The most important predictor of affiliation to this "high-risk group" was - as expected - the IRES summary score at admission. But unexpectedly influential were also some characteristics of social status such as lower household income and lower degrees of education. In rehabilitation research, analyses of change between pre- and post-measurement values should be accompanied by assessments of severity of rehabilitation status at discharge because even good improvements do not necessarily mean that a patient has been rehabilitated successfully.