Antarctic Base personnel live for 3 months in winter with no natural sunlight. This project compared sleep, by actigraphy, during periods of increased exposure to white light or blue enriched light in 2003. The primary aim was to help define the optimum spectral composition and intensity of artificial environmental light. Nine men and one woman (33 +/- 7 years, mean +/- SD), wore activity and light monitors continuously from 28.2 to 9.10, and kept sleep diaries. Extra light was provided by light boxes (standard white, 5300 K, or prototype blue enriched, 10 000K, Philips Lighting), which were turned on in bedrooms and in communal/work areas approximately 08.00-18.00 hours. After a no-treatment control period, 28.2-20.3, sequential 4-5 week periods of first white, then blue light, were imposed with a further control period 19.9-9.10. A limited baseline study in 2002 (no interventions) similarly measured light and activity in seven men and one woman (30 +/- 7 years). Daily light exposure in winter (lux, mean +/- SD) was doubled in 2003 (maximum 1039 +/- 281, average 64 +/- 21), compared to 2002 (572 +/- 276 and 30 +/- 11), P < 0.05 and P < 0.01, with no differences between white and blue light. There were no major differences in sleep between light conditions in 2003. A delay in sleep timing was found in midwinter compared to control (2003, bedtime, P < 0.05, sleep start, P < 0.05, sleep end, P < 0.01) and sleep fragmentation increased (P < 0.05). Sleep efficiency was slightly higher during all blue light periods compared to all white periods (P < 0.05). The use of higher intensity light of suitable spectral composition is proposed.