BACKGROUND: Due to long-term capacity problems in the psychiatric acute ward, we tried to canalise acute admissions due to life crises (and not serious mental disease) to a new short-term in-patient crisis unit. Our hypothesis was that the opening of this unit would lead to fewer admissions to the psychiatric acute ward and that this change would be reflected by an increase of patients with a more severe psychopathology. MATERIAL AND METHODS: The study had a quasi-experimental design. Two patient groups in a psychiatric acute ward (from separate catchment areas) were compared before (2.1.2003-1.6.2003) and after (2.1.2004-1.6.2004) establishment of a community based short-term inpatient crisis unit in one of the catchment areas. RESULTS: 234 patients were included in the study. Admissions to the psychiatric acute ward did not decline from any of the catchment areas from the first to the second time-period . The second time-period was associated with less psychopathology, but only for men in the area with a crisis unit. The reduction was largest for self-harm and suicidal behaviour (p = 0.02) and depression (p = 0.01). INTERPRETATION: None of our hypotheses were confirmed. Our main conclusion is that patient flow in acute mental health services involves a multitude of complex and unpredictable factors. The services continuously reorganise. Different ways of organising mental health services are rarely studied systematically, and such studies are difficult and resource demanding.