Background: The lives of adolescents in low and middle income countries would be improved by countries attaining the multifaceted Sustainable Development Goals (SDGs). For the most vulnerable young people this presents a major challenge. In response, the UNDP has proposed a new concept of 'development accelerators': provisions that simultaneously impact multiple SDGs. These may represent an opportunity for policy and provision to maximise the return on health and social investments. This study is the first to empirically test this concept, using a highly vulnerable sample of adolescents living with HIV in South Africa. Methods: Longitudinal data from 1060 adolescents living with HIV were gathered using community tracing from clinical records in 52 government clinics (90·1% reached, 93·5% 18-month retention). Standardised measures recorded eleven SDG-aligned targets across four goals, six potential accelerator provisions, and nine socio-demographic and HIV-related covariates. Associations of provisions with SDG outcomes were established by multivariable logistic regressions controlling for baseline outcomes, with adjustment for multiple outcome testing, and potential cumulative effects were tested by marginal effects modelling. Findings: Three provisions were 'development accelerators': each associated with positive outcomes across at least three SDGs: schools without physical violence ('safe schools'), government cash transfers and high parental monitoring (all p<0·05). For 9 out of 11 SDG-aligned targets, a combination of two of these accelerators showed cumulative positive associations, suggesting 'super-accelerator' effects of combination provisions. Interpretation: Findings suggest the validity of UNDP's 'accelerator' concept, identifying and providing services with multi-SDG impacts; and extend it to super-accelerators, that have cumulative effects. For adolescent sub-populations at elevated risk, development accelerators may provide a pathway to achieving health-focused SDG targets, and simultaneously impacting multiple SDG goals. Funding Statement: Nuffield Foundation, UKAID, Janssen Pharmaceutica, International AIDS Society, John Fell Fund, Economic and Social Research Council, European Research Council, Philip Leverhulme Trust, UNICEF. Declaration of Interests: LC, MO, LCa, ET, LS, AC and DW declare no conflicts of interest. LS reports grants from Sweden/Norad, grants from NIH, grants from MRC, personal fees from Gilead, other from AIDSImpact, editorial fees from Taylor and Francis, outside the submitted work. LC reports grants from Nuffield Foundation, UKAID and Sweden-Norad, Janssen Pharmaceutica South Africa, International AIDS Society, UNICEF Eastern and Southern Africa, John Fell Fund, European Research Council, Philip Leverhulme Trust, Economic and Social Research Council during the conduct of the study. Dr Webb is an employee of the UN Development Programme. Ethics Approval Statement: Ethical approval was given by the University of Cape Town (CSSR 2013/4), Oxford University (CUREC2/12-21), Provincial Departments of Health and Education and participating hospitals. All adolescents and their primary caregivers gave written informed consent at both time points in their language of choice (Xhosa or English), and read aloud in cases of low literacy.