OBJECTIVE: To assess follow-up and programmatic outcomes of HIV-exposed infants at Martin Preuss Centre, Lilongwe, from 2012 to 2014. METHODS: Retrospective cohort study using routinely collected HIV-exposed infant data. Data were analysed using frequencies and percentages in Stata v.13. RESULTS: Of 1035 HIV-exposed infants registered 2012-2014, 79% were available to be tested for HIV and 76% were HIV-tested either with DNA-PCR or rapid HIV test serology by 24 months of age. Sixty-five infants were found to be HIV-positive and 43% were started on antiretroviral therapy (ART) at different ages from 6 weeks to 24 months. Overall, 48% of HIV-exposed infants were declared lost-to-follow-up in the database. Of these, 69% were listed for tracing; of these, 78% were confirmed as lost-to-follow-up through patient charts; of these, 51% were traced; and of these, 62% were truly not in care, the remainder being wrongly classified. Commonest reasons for being truly not in care were mother/guardian unavailability to bring infants to Martin Preuss Centre, forgetting clinic appointments and transport expenses. Of these 86 patients, 36% were successfully brought back to care and 64% remained lost-to-follow-up. CONCLUSION: Loss to follow-up remains a huge challenge in the care of HIV-exposed infants. Active tracing facilitates the return of some of these infants to care. However, programmatic data documentation must be urgently improved to better follow-up and link HIV-positive children to ART.