In prevention of mother to child HIV transmission programs, nurses are responsible for counseling HIV-infected mothers about infant feeding. Studies of nurses' infant feeding counseling in African countries suggest that counseling quality is often poor and may be associated with nursing culture, inadequate training, and health system factors. Few studies, however, have conducted theory-informed process evaluations to assess nurses' implementation of an infant feeding counseling protocol. This study conducted the process evaluation of an infant feeding counseling protocol for the Breastfeeding, Antiretroviral and Nutrition (BAN) Study, an ongoing clinical trial in Lilongwe, Malawi. Six nurses, trained on World Health Organization recommendations, counseled HIV-infected mothers to exclusively breastfeed for 6 months and stop breastfeeding at 6 months. This study adapted and applied patient-provider communication theoretical concepts to assess nurses' counseling behaviors through: (1) direct observation and audio-taping of 123 infant feeding counseling sessions (30 antenatal and 93 postnatal) using a checklist to document nurses' adherence to specific protocol items; and (2) in-depth interviews with each nurse to document adherence perceptions and attitudes toward the protocol. Analysis included calculating a percent adherence of checklists and constant comparison techniques and visual display matrices for the observation and interview data. Results indicate that nurses were implementing the protocol at an average adherence level of 90% or above. Adherence to protocol items during the antenatal sessions was 100%. During postnatal sessions, adherence ranged from 92.7% to 97.8%. Nurses' implementation inconsistencies occurred when addressing breast health and breastfeeding cessation issues. An unexpected finding was nurses counseling mothers on proper infant formula preparation, although not detailed in the protocol. Interviews revealed a high nurse to mother ratio and additional counseling training on infant formula and complementary feeding preparation as perceived barriers toward complete protocol adherence. Findings illustrate that with minimal training, nurses implemented the BAN Study infant feeding counseling protocol as designed. The process evaluation results will help to interpret the BAN Study's impact and health outcomes related to evaluating mother's feasibility to follow infant feeding recommendations. The implication for HIV/AIDS clinical trials is to conduct process evaluations to identify and address potential implementation inconsistencies and, thereby, improve protocol implementation adherence.