Study objective We investigated two peer distribution models of HIV self-testing (HIVST) in HIV prevention demand creation compared with trained young community members (peer navigators).Methods We used restricted randomisation to allocate 24 peer navigator pairs (clusters) in KwaZulu-Natal 1:1:1: (1) standard of care (SOC): peer navigators distributed clinic referrals, pre-exposure prophylaxis (PrEP) and antiretroviral therapy (ART) information to 18–30 year olds. (2) peer navigator direct distribution (PND): Peer navigators distributed HIVST packs (SOC plus two OraQuick HIVST kits) (3) incentivised peer networks (IPN): peer navigators recruited young community members (seeds) to distribute up to five HIVST packs to 18–30 year olds within their social networks. Seeds received 20 Rand (US$1.5) for each recipient who distributed further packs. The primary outcome was PrEP/ART linkage, defined as screening for PrEP/ART eligibility within 90 days of pack distribution per peer navigator month (pnm) of outreach, in women aged 18–24 (a priority for HIV prevention). Investigators and statisticians were blinded to allocation. Analysis was intention to treat. Total and unit costs were collected prospectively.Results Between March and December 2019, 4163 packs (1098 SOC, 1480 PND, 1585 IPN) were distributed across 24 clusters. During 144 pnm, 272 18–30 year olds linked to PrEP/ART (1.9/pnm). Linkage rates for 18–24-year-old women were lower for IPN (n=26, 0.54/pnm) than PND (n=45, 0.80/pnm; SOC n=49, 0.85/pnm). Rate ratios were 0.68 (95% CI 0.28 to 1.66) for IPN versus PND, 0.64 (95% CI 0.26 to 1.62) for IPN versus SOC and 0.95 (95% CI 0.38 to 2.36) for PND versus SOC. In 18–30 year olds, PND had significantly more linkages than IPN (2.11 vs 0.88/pnm, RR 0.42, 95% CI 0.18 to 0.98). Cost per pack distributed was cheapest for IPN (US$36) c.f. SOC (US$64). Cost per person linked to PrEP/ART was cheaper in both peer navigator arms compared with IPN.Discussion HIVST did not increase demand for PrEP/ART. Incentivised social network distribution reached large numbers with HIVST but resulted in fewer linkages compared with PrEP/ART promotion by peer navigators.Trial registration number NCT03751826.Data are available upon request. Data can be can access and download the analytical datasets through the AHRI data repository: https://data.africacentre.ac.za. https://data.ahri.org/index.php/home