In plain terms: Bright Light Therapy for Depression has some human trial evidence supporting it, but the evidence is not yet strong enough for EBL to call it established. It is worth knowing about and worth discussing with your clinician.
Bright Light Therapy for Depression
The evidence as the record states it
Winter seasonal depression. Terman et al. (Arch Gen Psychiatry, 1998; PMID 9783557): 158 patients, 10,000 lux for 30 minutes; morning and evening light did not differ on the main analysis and both beat a low-density negative ion placebo; morning light did better only on a stringent remission criterion; high-density negative ions also worked; NIMH-funded. Eastman et al. (Arch Gen Psychiatry, 1998; PMID 9783558): 96 patients, about 6,000 lux for 90 minutes a day for 4 weeks against a sham negative ion generator; no difference in mean depression scores at 4 weeks, but more full or near-full remissions on morning light from week 3 (61% morning, 50% evening, 32% placebo by one criterion); NIMH-funded. Lewy et al. (Arch Gen Psychiatry, 1998; PMID 9783559): 51 patients, morning light shifted the body clock earlier and was more antidepressant than evening light; no placebo arm; NIH-funded. Lam et al. (Am J Psychiatry, 2006; PMID 16648320): 96 patients, 10,000 lux light plus placebo capsule against dim light plus fluoxetine 20 mg for 8 weeks; no difference between them, response 67% in each group; no double-placebo arm. Golden et al. (Am J Psychiatry, 2005; PMID 15800134): meta-analysis, bright light effect size 0.84 (8 trials) and dawn simulation 0.73 (5 trials) in seasonal depression. Chen et al. (J Affect Disord, 2024; PMID 38220102): network meta-analysis of 21 trials and 1,037 people; light beat control treatments, with the quality of evidence rated low. Wan et al. (Medicine, 2025; PMID 40629637): network meta-analysis of 17 trials and 773 patients; white light ranked above green, blue and red. Non-seasonal major depression. Lam et al. (JAMA Psychiatry, 2016; PMID 26580307): 122 adults, double-blind, 8 weeks, 10,000 lux for 30 minutes early in the morning; on the MADRS, light alone beat an inactive negative ion generator plus placebo pill (effect size 0.80) and light plus fluoxetine did best (1.11), while fluoxetine alone did not separate from placebo (0.24); funded by the Canadian Institutes of Health Research. Menegaz de Almeida et al. (JAMA Psychiatry, 2025; PMID 39356500): 11 trials and 858 patients, light added to treatment; remission 40.7% versus 23.5% (odds ratio 2.42) and response 60.4% versus 38.6% (odds ratio 2.34). Tong et al. (Asian J Psychiatr, 2024; PMID 39067131): 15 trials and 883 patients; standardized mean difference 0.48. Wang et al. (Gen Hosp Psychiatry, 2026; PMID 42097096): trials from 2020 to 2025; standardized mean difference 0.35 in non-seasonal major depression. Tuunainen et al. (Cochrane, 2004; PMID 15106233): 20 trials, mostly light added to drugs or sleep deprivation and mostly under 8 days; the overall benefit did not reach statistical significance, while high-quality trials (standardized mean difference 0.90) and morning-light trials did; hypomania was more common on bright light (risk ratio 4.91, number needed to harm 8). Kept in the record: Lam et al. (Can J Psychiatry, 2020; PMID 31826657), 7 trials and 259 patients with bipolar depression, light added to treatment improved Hamilton scores (standardized mean difference 0.43) with no difference in mood switches, evidence the authors call nonconclusive; Nussbaumer-Streit et al. (Cochrane, 2019; PMID 30883670), light to prevent winter depression, one eligible trial of 46 people, very low quality, no conclusion possible.
Why Grade B and not Grade A
Grade B reflects moderate human evidence -- typically a single strong trial, multiple smaller trials, or surrogate endpoint data without hard clinical outcome replication. To reach Grade A, Bright Light Therapy for Depression would require at least one additional independent randomized trial demonstrating a hard clinical endpoint or objectively measured functional outcome in a comparable population.
What this grade covers and what it does not
What would move this grade
An independent randomized trial demonstrating a hard clinical endpoint -- mortality, major adverse event, or objectively confirmed functional improvement -- in an adequately powered sample, with results independently replicated, would support Grade A.
How this was studied
Form: White light box, fluorescent in most trials, at about 10,000 lux at the eyes (about 6,000 lux in one trial); placebo a dim light or an inactive negative ion generator · Dose: 30 minutes a day at 10,000 lux, mostly early in the morning (90 minutes a day at 6,000 lux in one trial) · Population: Adults with winter seasonal depression, or with non-seasonal major depression of at least moderate severity; in the non-seasonal trials light was often added to an antidepressant · Duration: 1 to 8 weeks · What is sold differs: Light boxes, lamps and glasses sold for mood vary in brightness, spectrum and the distance at which the stated lux is reached. This grade describes white light at about 10,000 lux at the distance the trials used; blue-only, green and red light ranked below white light in a 2025 network meta-analysis (PMID 40629637).