The Intervention with Microfinance for AIDS and Gender Equity (IMAGE) combines microfinance, gender/HIV training and community mobilization (CM) in South Africa. A trial found reduced intimate partner violence among clients but less evidence for impact on sexual behaviour among clients' households or communities. This process evaluation examined how feasible IMAGE was to deliver and how accessible and acceptable it was to intended beneficiaries during a trial and subsequent scale-up. Data came from attendance registers, financial records, observations, structured questionnaires (378) and focus group discussions and interviews (128) with clients and staff. Gender/HIV training and CM were managed initially by an academic unit ('linked' model) and later by the microfinance institution (MFI) ('parallel' model). Microfinance and gender/HIV training were feasible to deliver and accessible and acceptable to most clients. Though participation in CM was high for some clients, others experienced barriers to collective action, a finding which may help explain lack of intervention effects among household/community members. Delivery was feasible in the short term but both models were considered unsustainable in the longer term. A linked model involving a MFI and a non-academic partner agency may be more sustainable and is being tried. Feasible models for delivering microfinance and health promotion require further investigation.