Introduction Pre-exposure prophylaxis (PrEP) is a highly effective HIV prevention strategy and is recommended for populations at risk of HIV, including adolescent girls and young women (AGYW) in HIV endemic settings. However, PrEP continuation and high adherence remain challenges to its impact. Existing PrEP adherence interventions can be time- and cost-intensive. Widescale PrEP delivery will require the identification of layered PrEP support strategies for AGYW with diverse prevention needs. We describe the design of a sequential multiple assignment randomized trial (SMART) to evaluate a PrEP adherence support model using scalable, stepped interventions in AGYW in South Africa. Methods “PrEP SMART” is a randomized trial in Johannesburg, South Africa, enrolling AGYW who are between 18 and 25 years of age, sexually active, newly initiating PrEP, and have regular access to a mobile phone. Participants are randomized 1:1 to standard-of-care PrEP counseling with either two-way SMS or WhatsApp group adherence support. Adherence is assessed at three months using tenofovir diphosphate (TFV-DP) levels from dried blood spots collected at month 2 to categorize participants as “responders” (TFV-DP ≥500 fmol/punch) or “non-responders” (TFV-DP <500 fmol/punch). AGYW defined as ‘non-responders’ undergo a secondary 1:1 randomization to either quarterly drug-level feedback counseling or monthly issue-focused counseling, in addition to their first-level intervention. The primary outcome is PrEP adherence at nine months (TFV-DP ≥700 fmol/punch). We will assess the effect of our two initial interventions on TFV-DP levels among responders, assess the effect of our intensified interventions on TFV-DP levels among non-responders, and identify the optimal sequence of adherence interventions through nine months. Trial registration ClinicalTrials.gov, NCT04038060. Registered on 30 July 2019.