The Kgolo Mmogo study is a randomised controlled intervention trial that examines the effectiveness of a group intervention to enhance resilience in HIV-infected South African mothers (N = 427) and their young children (N = 435). We describe here how the severity of psychological and social problems experienced by some of the study participants required referrals for other services and discuss the barriers encountered in facilitating such referrals. Over a 30-month period 54 mothers and 59 children were referred for additional support. For mothers, the most frequent reasons for referral related to domestic violence and problems within relationships, while for children the most common grounds for referral were the evaluation and treatment of behavioural problems and severe emotional disturbances, including depression. Eight children were referred for suspected abuse. Observations from the study demonstrate that current systems for referral are overloaded and that there is a paucity of specialised services available. Our experience suggests that participants may benefit from using the intervention as a first point of support and that psychosocial referrals should perhaps be delayed until functional advice is provided (within the group) on ways of accessing wider support effectively. The intervention may also benefit from the inclusion of an intervention team member who is specifically tasked to follow up on referrals. This includes follow-up for participants who were not included in the group intervention. Furthermore, we argue that socio-economic constraints, which often manifest as lack of mobility to access service delivery, can severely impact on the implementation of an intervention study in a developing context. This constraint is experienced in terms of limited access to experimental intervention groups and services from referrals.