BACKGROUND: Index-linked HIV testing, whereby children of individuals with HIV are targeted for testing, increases HIV yield but relies on uptake. Community-based testing might address barriers to testing access. In the Bridging the Gap in HIV testing and care for children in Zimbabwe (B-GAP) study, we investigated the uptake and yield of index-linked testing in children and the uptake of community-based vs facility-based HIV testing in Zimbabwe. METHODS: B-GAP was an interventional study done in the city of Bulawayo and the province of Matabeleland South between Jan 29 and Dec 12, 2018. All HIV-positive attendees (index patients) at six urban and three rural primary health-care clinics were offered facility-based or community-based HIV testing for children (age 2-18 years) living in their households who had never been tested or had tested as HIV-negative more than 6 months ago. Community-based options involved testing in the home by either a trained lay worker with a blood-based rapid diagnostic test (used in facility-based testing), or by the child's caregiver with an oral HIV test. Among consenting individuals, the primary outcome was testing uptake in terms of the proportion of eligible children tested. Secondary outcomes were uptake of the different HIV testing methods, HIV yield (proportion of eligible children who tested positive), and HIV prevalence (proportion of HIV-positive children among those tested). Logistic regression adjusting for within-index clustering was used to investigate index patient and child characteristics associated with testing uptake, and the uptake of community-based versus facility-based testing. FINDINGS: Overall, 2870 index patients were linked with 6062 eligible children (3115 [51·4%] girls [sex unknown in seven], median age 8 years [IQR 5-13]). Testing was accepted by index patients for 5326 (87·9%) children, and 3638 were tested with a known test outcome, giving an overall testing uptake among 6062 eligible children of 60·0%. 39 children tested positive for HIV, giving an HIV prevalence among the 3638 children of 1·1% and an HIV yield among 6062 eligible children of 0·6%. Uptake was positively associated with female sex in the index patient (adjusted odds ratio [aOR] 1·56 [95% CI 1·38-1·77], p<0·0001) and child (aOR 1·10 [1·03-1·19], p=0·0080), and negatively associated with any financial cost of travel to a clinic (aOR 0·86 [0·83-0·88], p<0·0001), increased child age (6-9 years: aOR 0·99 (0·89-1·09); 10-15 years: aOR 0·91 [0·83-1·00]; and 16-18 years: aOR 0·75 [0·66-0·85]; p=0·0001 vs 2-5 years), and unknown HIV status of the mother (aOR 0·81 [0·68-0·98], p=0·027 vs HIV-positive status). Additionally, children had increased odds of being tested if community-based testing was chosen over facility-based testing at screening (1320 [73·9%] children tested of 1787 vs 2318 [65·5%] of 3539; aOR 1·49 [1·22-1·81], p=0·0001). INTERPRETATION: The HIV yield of index-linked testing was low compared with blanket testing approaches in similar settings. Index-linked HIV testing can improve testing uptake among children, although strategies that improve testing uptake in older children are needed. Community based testing by lay workers is a feasible strategy that can be used to improve uptake of HTS among children and adolescents. FUNDING: UK Medical Research Council, UK Department for International Development, Wellcome Trust.