Introduction The prevalence of type 2 diabetes mellitus (T2DM) and associated morbidity and mortality are increasing in sub-Saharan Africa (SSA). To facilitate access to quality care and improve treatment outcomes, there is a need for innovative community care models and optimized use of non-physician healthcare workers bringing diagnosis and care closer to patients’ homes. Aim We aimed to describe with a scoping review different models of community-based care for non-pregnant adults with T2DM in SSA, and to synthesize the model outcomes in terms of engagement in care, blood sugar control, acceptability, and end-organ damage. We further aimed to critically appraise the different models of care and compare community-based to facility-based care if data were available. Methods We searched Medline, Embase, Cumulative Index to Nursing and Allied Health Literature (CINAHL) and Scopus, supplemented with backward and forward citation searches. We included cohort studies, randomized trials and case-control studies that reported on non-pregnant individuals diagnosed with T2DM in SSA, who received a substantial part of care in the community. Only studies which reported at least one of our outcomes of interest were included. A narrative analysis was conducted, and comparisons made between community-based and facility-based models, where within-study comparison was reported. Results 5,335 unique studies were retrieved, four of which met our inclusion criteria. Most studies were excluded because interventions were facility-based; community care interventions described in the studies were add-on features of a primarily facility-based care; and studies did not report outcomes of interest. The included studies reported on a total of 383 individuals with T2DM. Three different community care models were identified. 1) A community-initiated model where diagnosis, treatment and monitoring occurred primarily in the community. This model reported a higher linkage and engagement in care at 9 months compared to the corresponding facility model, but only slight reductions of average blood glucose levels at six months compared to baseline. 2) A facility-originated community model where after treatment initiation, a substantial part of follow-up was offered at community level. Two studies reported such a model of care, both had as core component home-delivery of medication. Acceptability of this approach was high. But neither study found improved T2DM control when compared to facility care 3) An eHealth model with high acceptability scores for both patients and care providers, and an absolute 1.76% reduction in average HbA1c levels at two months compared to baseline. There were no reported outcomes on end-organ damage. All four studies were rated as being at high risk for bias. Conclusion Evidence on models of care for persons with T2DM in SSA where a substantial part of care is shifted to the community is scant. Whereas available literature indicates high acceptability of community-based care, we found no conclusive data on their effectiveness in controlling blood sugar and preventing complications. Evidence from larger scale studies, ideally randomized trials with clinically relevant endpoints is needed before roll-out of community-based T2DM care can be recommended in SSA.