Depression occurs in about 10% of pregnant women and is difficult to treat because of the need to consider safety of mother and child. Our aim is to study the clinical efficacy of light therapy in depression during pregnancy, as a non-drug biological treatment with few side effects.New studies point to increased long-term developmental risks in babies following antidepressant drugs during pregnancy. A treatment of depression during pregnancy that is efficacious, reliable, and safe, is an urgent unmet clinical need for both women and their psychiatrists. Promising pilot studies of light treatment inspired us to begin this controlled trial. The specific aims are to compare clinical improvement between two light intensity groups (active vs placebo), and to test whether response is related to changes in biological clock timing. The most apparent potential outcome of the study is the reduction of depression in pregnant women. If proven effective, light therapy will be a treatment with minimal, if any, risk to the growing foetus. Another benefit is that light provides a biological treatment with minor side effects compared to antidepressants. Many women also prefer such a "natural" treatment rather than medication during their pregnancy. Light therapy for winter depression is well established in Switzerland and reimbursed by medical insurance. This means that a new indication could rapidly and inexpensively be implemented in the community. The study is a double-blind randomised clinical trial lasting five weeks, with a parallel design using 7000 lux (active) or 70 lux (placebo) light boxes for one hour per day on awakening. We measure rest-activity cycles over 24 hours with actimeters that also measure light exposure. Evening saliva samples are collected before and after 5 weeks to measure the circadian rhythm of melatonin secretion, a validated marker of the biological clock. Recruitment of participants is difficult, because many women are not prepared to invest in the time needed to answer questionnaires and have interviews. So far we have an overall response rate of 60% (we do not know which patient received the bright light until the end of the study).