Introduction: Total sleep deprivation (TSD) has acute antidepressive properties, however, most of the TSD responders relapse immediately after recovery sleep. Even short naps during the day after TSD may reverse the beneficial effects of the procedure, especially when scheduled in the morning hours.The impact of TSD as well as the impact of naps on the mood of depressed patients demonstrate a high chronobiological dependence: Partial sleep deprivation in the second half of the night has a better influence on mood than in the first half of the night and naps in the morning hours have a more detrimental effect on mood than naps in the afternoon. In addition, Wehr et al. (1979) advanced the sleep period for 6 hours, which resulted in an improvement of depressed patients after 2 weeks. The authors recommended avoiding sleep during the early morning hours in depression and hypothesized a critical phase in these hours because of the depressiogenic effect of sleep during this time.In a pilot study we were able to demonstrate, that the improvement of mood due to TSD can be stabilized by a succeeding phase advance of the sleep period (SPA) in the following days. After the TSD patients slept the next day from 17.00-24.00 hrs and then bedtime was shifted back by one hour daily until the normal bedtime from 23.00-6.00 hrs was reached seven days later. This method prevented about 60% of the TSD-responders from relapsing into depression and the effect was independent of concommitant medication.To test the efficacy of TSD + SPA in therapy-refractory depression (TRD), age- and gender-matched subgroups with or without TRD were drawn from a larger sample of depressed inpatients who participated in our study.Methods: TRD was defined as non-response to two antidepressant drugs, which had been administered for 4-6 weeks each. Major depressive disorder was diagnosed according to DSM-IV criteria. TSD- and SPA-response was defined as at least 30% improvement on the 6-HAMD.12 TRD-patients (6 males, 6 females; 49.3 ± 12.2 years) who responded to TSD out of a total sample of 66 TSD-responder were selected. Mean duration of the current episode was 59.0 ± 72.4 weeks, the mean 21-Item Hamilton Rating Scale (21-HAMD) attained 24.5 ± 7.5 before TSD.These patients were compared with 12 TSD-responders (8 males, 4 females; 44.8 ± 8.6 years) who did not fullfill the TRD criteria. Mean duration of the current episode was 12.9 ± 8.8 weeks, the 21-HAMD reached 28.1 ± 4.8. Ten of the TRD-patients and 9 of the comparison group had diurnal variations of mood. Depressed mood was measured daily by the 6-HAMD (a modified version of the 21-HAMD).Results: Ten of 12 TRD-patients and 8 of 12 patients of the control group were SPA-responders. The 21-HAMD after SPA-treatment was 12.1 ± 8.9 in the TRD-group and 11.1 ± 6.9 in the control-group. The SPA-response in the TRD-group was 55.6% ± 29.3 concerning the 21-HAMD and reached 51.6% ± 35.8 in the control-group. Both groups did not differ concerning psychopathological or demographic data, except the mean duration of the current episode (p = 0.040).Conclusion: Our data demonstrate a clinically relevant effect of TSD combined with SPA in therapy refractory depression. The TRD-group with a significant longer mean duration of the current episode showed an encouraging stabilization of mood after successful TSD when subjected to succeeding SPA. Insofar, patients with TRD should be motivated to participate in TSD + SPA, as this method seems to provide a high success rate. Prospective studies are warranted to confirm our observations.