Background: In 1984, Rosenthal et al. described a group of patients characterised by repeated winter depression with atypical symptoms eventually followed by summer mania or hypomania (seasonal affective disorder, SAD). The relationship between SAD and the classical manic-depressive disorder is uncertain. The aim of this study was to validate the Seasonal Pattern Assessment Questionnaire (SPAQ) classification in relation to the DSM-III-R criteria of seasonal patterns in a group of patients with bipolar affective disorder and to evaluate the stability of the SPAQ score index over time. Methods: A group of bipolar patients were followed for 3 years with examinations every third month and at hospitalisation. At inclusion and at each following examination the patients were rated with the Hamilton Depression Rating Scale, the Newcastle Depression Rating Scale, and the Bech-Rafaelsen Mania Rating Scale. At inclusion and once a year the patients completed the SPAQ. Results: Fifty-six patients agreed to participate and 46 patients completed 1 year or more. Eight patients, all women, had at least one SPAQ score index (SSI) of 11 or more. Women scored significantly higher than men on SPAQ. The test/re-test reliability of SSI between two consecutive years was good, but decreased as the time difference between tests increased. Moreover, we found no relation between seasonal variations of affective episodes according to SSI and DSM-III-R. Limitations: The group of patients with seasonality according to DSM-III-R or SSI was small. Conclusion: The validity of SSI between two consecutive years is good, but decreases as the time difference between tests increases. There is no relationship between seasonality defined by DSM-III-R and SSI. Female bipolar patients show, as the general population, seasonal variations in mood, energy, sociality, appetite, and sleep independently of their affective episodes. SSI cannot be used for the prediction of seasonal variation in a group of bipolar patients.