The observation of a progressive recurrence in affective disorder has been interpreted as a process of sensitisation. The clinical applicability of such a theoretical model was investigated using the Danish case register, which includes all hospital admissions with primary affective disorder in Denmark from 1971 to 1993. A total of 8,737 patients admitted to a psychiatric hospital at least twice constituted the study sample. Information on treatment intervention was not available. Measures describing the initial course of admission episodes were defined in three different ways: 1) a short period between initial episodes 2) decreasing intervals between initial episodes or 3) a combination of 1) and 2). Socio-demographic variables such as gender, age at onset and marital status differentiated between the three types of measures and the measures also demonstrated different effects in predicting the risk of further recurrence. In unipolar disorder, patients with a decreasing interval between episodes had the greatest risk of further recurrence, whereas for bipolar patients, a short period between episodes played a more important role than the sequence of episodes in itself.