Integrating voluntary HIV counselling and testing (VCT) with family planning and other reproductive health services may be one effective strategy for expanding VCT service delivery in resource poor settings. Using 30,257 VCT client records with linked facility characteristics from Ethiopian non-governmental, non-profit, reproductive health clinics, we constructed multi-level logistic regression models to examine associations between HIV and family planning service integration modality and three outcomes: VCT client composition, client-initiated HIV testing and client HIV status. Associations between facility HIV and family planning integration level and the likelihood of VCT clients being atypical family planning client-types, versus older (at least 25 years old), ever-married women were assessed. Relative to facilities co-locating services in the same compound, those offering family planning and HIV services in the same rooms were 2–13 times more likely to serve atypical family planning client-types than older, ever-married women. Facilities where counsellors jointly offered HIV and family planning services and served many repeat family planning clients were significantly less likely to serve single clients relative to older, married women. Younger, single men and older, married women were most likely to self-initiate HIV testing (78.2 and 80.6% respectively), while the highest HIV prevalence was seen among older, married men and women (20.5 and 34.2% respectively). Compared with facilities offering co-located services, those integrating services at room- and counsellor-levels were 1.9–7.2 times more likely to serve clients initiating HIV testing. These health facilities attract both standard material and child health (MCH) clients, who are at high risk for HIV in these data, and young, single people to VCT. This analysis suggests that client types may be differentially attracted to these facilities depending on service integration modality and other facility-level characteristics.