BACKGROUND: The HPTN 071 (PopART) cluster-randomised trial provided door-to-door HIV testing services to a large proportion of individuals residing in 21 intervention communities in Zambia and South Africa from 2014 to 2017 and reached the UNAIDS first 90 target among women in Zambia, yet gaps remained among men and young adults. This cluster-randomised study nested in the HPTN 071 (PopART) trial sought to increase knowledge of HIV status across all groups by offering the choice of oral HIV self-testing in addition to routine door-to-door HIV testing services. METHODS: We nested this cluster-randomised trial in four HTPN 071 (PopART) intervention communities in northern Zambia. 66 zones (clusters) in these communities were randomly allocated (1:1) to either oral HIV self-testing plus routine door-to-door HIV testing services (HIV self-testing group) or the PopART standard of care of door-to-door HIV testing services alone (non- HIV self-testing group) over a 3-month period. All individuals aged 16 years or older were eligible for HIV testing. Randomisation was achieved by randomly selecting one allocation from a list of 10 000 possible allocations during a public ceremony. In HIV self-testing zones, trained lay-counsellors (known as community HIV care providers) visited households and offered eligible individuals the choice of HIV testing using HIV self-testing or routine door-to-door HIV testing services. For individuals aged 18 years or older whose partner was absent during the household visit, an HIV self-test kit could be left for secondary distribution to the absent partner. The primary outcome was knowledge of HIV status (defined as self-reporting HIV positive to the community HIV care providers or accepting an offer of HIV testing services). Outcomes were measured among households that were first visited, and individuals first enumerated as a household member during the HIV self-testing intervention period. We analysed data at the individual level using population-average logistic regression models, accounting for clustering of outcomes by zone, to estimate the effect of the intervention. This trial is registered with ClinicalTrials.gov, number NCT02994329. FINDINGS: Between Feb 1, and April 30, 2017, the community HIV care providers enumerated 13 267 eligible individuals in the HIV self-testing group and 13 706 in the non-HIV self-testing group. After intervention implementation, 9027 (68%) of 13 267 in the HIV self-testing group had knowledge of HIV status compared with 8952 (65%) of 13 706 in the non-HIV self-testing group (adjusted odds ratio 1·30, 95% CI 1·03-1·65; p=0·03). The effect differed by sex (p=0·01). Among men, knowledge of HIV status was higher in the HIV self-testing group than in the non-HIV self-testing group (3843 [60%] of 6368 vs 3571 [55%] of 6486; adjusted odds ratio 1·31, 95% CI 1·07-1·60; p=0·01). There was no evidence of a between-group difference among female participants. INTERPRETATION: Providing a choice of HIV self-testing during delivery of door-to-door HIV testing services increased knowledge of HIV status, driven by an effect among men. Lay counsellors have a vital role to play in adapting HIV self-testing interventions to local context. FUNDING: The International Initiative for Impact Evaluation (3ie), the Bill & Melinda Gates Foundation, National Institute of Allergy and Infectious Diseases, National Institute on Drug Abuse, National Institute of Mental Health, and the President's Emergency Plan for AIDS Relief.