BACKGROUND: The recent development of rapid diagnostic tests (RDTs) for human African trypanosomiasis (HAT) enables elimination programmes to decentralise serological screening services to frontline health facilities. However, patients must still undertake multiple onwards referral steps to either be confirmed or discounted as cases. Accurate surveillance thus relies not only on the performance of diagnostic technologies but also on referral support structures and patient decisions. This study explored why some RDT-positive suspects failed to complete the diagnostic referral process in West Nile, Uganda. METHODS: Between August 2013 and June 2015, 85% (295/346) people who screened RDT-positive were examined by microscopy at least once; 10 cases were detected. We interviewed 20 RDT-positive suspects who had not completed referral (16 who had not presented for their first microscopy examination, and 4 who had not returned for a second to dismiss them as cases after receiving discordant [RDT-positive, but microscopy-negative results]). Interviews were analysed thematically to examine experiences of each step of the referral process. RESULTS: Poor provider communication about HAT RDT results helped explain non-completion of referrals in our sample. Most patients were unaware they were tested for HAT until receiving results, and some did not know they had screened positive. While HAT testing and treatment is free, anticipated costs for transportation and ancillary health services fees deterred many. Most expected a positive RDT result would lead to HAT treatment. RDT results that failed to provide a definitive diagnosis without further testing led some to question the expertise of health workers. For the four individuals who missed their second examination, complying with repeat referral requests was less attractive when no alternative diagnostic advice or treatment was given. CONCLUSIONS: An RDT-based surveillance strategy that relies on referral through all levels of the health system is inevitably subject to its limitations. In Uganda, a key structural weakness was poor provider communication about the possibility of discordant HAT test results, which is the most common outcome for serological RDT suspects in a HAT elimination programme. Patient misunderstanding of referral rationale risks harming trust in the whole system and should be addressed in elimination programmes.