A high-intensity fluorescent lighting system, tilted downward toward the head, and emitting negligible levels of ultraviolet radiation, was tested under two random crossover protocols in winter-depressed patients: 30-minute sessions at (a) 3,000 lux vs. 10,000 lux in early morning, and (b) morning vs. evening sessions at 10,000 lux. Judgment of clinical remission was based jointly on relative and absolute score improvements on a Structured Interview Guide for the Hamilton Depression Scale--Seasonal Affective Disorder Version (SIGH-SAD) and a set of supplementary atypical-vegetative items. Data are presented for 24 subjects who showed relapse upon withdrawal. An overall remission rate of 75 percent was found for morning light at 10,000 lux. The rates for evening light (25%) and 3,000 lux morning light (19%) were significantly lower. The remission rate for morning light treatment of 10,000 lux for 30 minutes approximately equalled 2,500 lux treatment for 2 hours (data from our earlier studies), suggesting a reciprocity between dosing dimensions of intensity and duration. No pathological changes were revealed by ophthalmological examinations given after 2 to 6 weeks of daily treatment.