Background: Depression in women is one of the commonest problems encountered in primary care. We aimed to compare the effectiveness of a stepped-care programme with usual care in primary-care management of depression in low-income women in Santiago, Chile. Methods: In a randomised controlled trial, in three primary-care clinics in Chile, 240 adult female primary-care patients with major depression were allocated stepped care or usual care. Stepped care was a 3-month, multicomponent intervention led by a non-medical health worker, which included a psychoeducational group intervention, structured and systematic follow-up, and drug treatment for patients with severe depression. Data were analysed on an intention-to-treat basis. The primary outcome measure was the Hamilton depression rating scale (HDRS) administered at baseline and at 3 and 6 months after randomisation. Findings: About 90% of randomised patients completed outcome assessments. There was a substantial between-group difference in all outcome measures in favour of the stepped-care programme. The adjusted difference in mean HDRS score between the groups was -8.89 (95% CI -11.15 to -6.76; p<0.0001). At 6-months' follow-up, 70% (60-79) of the stepped-care compared with 30% (21-40) of the usual-care group had recovered (HDRS score <8). Interpretation: Despite few resources and marked deprivation, women with major depression responded well to a structured, stepped-care treatment programme, which is being introduced across Chile. Socially disadvantaged patients might gain the most from systematic improvements in treatment of depression.