Background HIV diagnosis is a critical step in linking HIV-infected individuals to care and treatment and linking HIV-uninfected persons to prevention services. However, the uptake of HIV testing remains low in many countries. HIV self-screening (HIVSS) is acceptable to adults, but there is limited data on HIVSS feasibility in community programmes. This study aimed to evaluate the feasibility of HIVSS in South Africa. Methods We conducted a prospective study that enrolled participants through mobile site, homebased, workplace and sex worker programmes in two townships from May to November 2017. Following an information session on HIVSS, interested participants were offered one of three methods of HIVSS testing: supervised, semi-supervised, and unsupervised. Participants who opted for unsupervised testing and those who tested HIV positive after semi- or supervised HIVSS were followed up telephonically or with a home visit one week after receipt of the test kit to confirm results and linkages to care. Follow-up visits were concluded when the participant indicated that they had used the kit or had accessed a confirmatory HIV test. Results Of the 2061 people approached, 88.2% (1818/2061) received HIV testing information. Of this group, 89% (1618/1818) were enrolled in the study and 70.0% (1133/1618) were tested for HIV with the kit. The median age was 28 (IQR:23–33) years with an even gender distribution. Of those enrolled, 43.0% (696/1618) were identified through homebased outreach, 42.5% (687/1618) through mobile sites, 7.3% (118/1618) at their workplace and 7.2% (117/1618) from sex worker programmes. A total of 68.7% (1110/1616) selected unsupervised HIVSS, whereas 6.3% (101/1616) opted for semi-supervised and 25.0% ((405/1616) chose supervised HIVSS. Overall, the HIV prevalence using the HIVSS test was 8.2% (93/1129). Of those newly diagnosed with HIV, 16% (12/75) were initiated on ART. Almost half (48.0%; 543/1131) of those tested were linked to a primary HIV test as follows: supervised (85.2%; 336/394); semi-supervised (93.8%; 91/97) and unsupervised (18.1%; 116/640). Conclusion Unsupervised HIVSS was by far the most selected and utilised HIVSS method. Linkages to primary and confirmatory testing for the unsupervised HIVSS and further care were low, despite home visits and telephonic reminders.