Background: Many migrants face adverse socioeconomic conditions and barriers to health services that can impair timely HIV diagnosis and access to life-saving treatments. We aimed to assess the differences in overall mortality by geographical origin in HIV-positive men and women using data from COHERE, a large European collaboration of HIV cohorts from 1997 to 2013. Methods: In this observational cohort study, we included HIV-positive, antiretroviral-naive people accessing care in western Europe from COHERE. Individuals were eligible if enrolled in a cohort that collected information on geographical origin or ethnic origin from Jan 1, 1997, to March 19, 2013, aged 18-75 years, they had available information about sex, they were not infected perinatally or after the receipt of clotting factor concentrates, and were naive to combination antiretroviral therapy at cohort entry. Migrants' origins were grouped into seven regions: western Europe and similar countries (Australia, Canada, New Zealand, and the USA); eastern Europe; North Africa and the Middle East; sub-Saharan Africa; Latin America; the Caribbean; and Asia and the rest of Oceania (excluding Australia and New Zealand). Crude and adjusted mortality rate ratios were calculated by use of Poisson regression stratified by sex, comparing each group with the native population. Multiple imputation with chained equations was used to account for missing values. Findings: Between Oct 25, 1979, and March 19, 2013, we recruited 279 659 individuals to the COHERE collaboration in EuroCoord. Of these 123 344 men and 45 877 women met the inclusion criteria. Our data suggested effect modification by transmission route (pinteraction=0·12 for men; pinteraction=0·002 for women). No significant difference in mortality was identified by geographical origin in men who have sex with men. In heterosexual populations, most migrant men had mortality lower than or equal to that of native men, whereas no group of migrant women had mortality lower than that in native women. High mortality was identified in heterosexual men from Latin America (rate ratio [RR] 1·46, 95% CI 1·00-2·12, p=0·049) and heterosexual women from the Caribbean (1·48, 1·29-1·70, p