OBJECTIVES: To evaluate the effect of storing commonly used rapid diagnostic tests above manufacturer-recommended temperature (at 37°C), and the accuracy of delayed reading of oral fluid kits with relevance to HIV self-testing programmes. DESIGN: A quality assurance study of OraQuick (OraSure), Determine HIV 1/2™ (Alere) and Uni-Gold™ (Recombigen®). METHODS: Consecutive adults (≥18y) attending Ndirande Health Centre in urban Blantyre, Malawi in January to April 2012 underwent HIV testing with two of each of the three rapid diagnostic test kits stored for 28 days at either 18°C (optimally-stored) or at 37°C (pre-incubated). Used OraQuick test kits were stored in a laboratory for delayed day 1 and subsequent monthly re-reading was undertaken for one year. RESULTS: Of 378 individuals who underwent parallel testing, 5 (1.3%) were dropped from the final analysis due to discordant or missing reference standard results (optimally-stored Determine and Uni-Gold). Compared to the diagnostic reference standard, OraQuick had a sensitivity of 97.2% (95% CI: 93.6-99.6). There were 7 false negative results among all test kits stored at 37°C and three false negatives among optimally stored kits. Excellent agreement between pre-incubated tests and optimally-stored tests with Kappa values of 1.00 for Determine and Uni-Gold; and 0.97 (95% CI: 0.95; 1.00) for OraQuick were observed. There was high visual stability on re-reading of OraQuick, with only 1/375 pre-incubated and 1/371 optimally-stored OraQuick kits changing from the initial result over 12 months. CONCLUSION: Erroneous results observed during HIV testing in low income settings are likely to be due to factors other than suboptimal storage conditions. Re-reading returned OraQuick kits may offer a convenient and accurate quality assurance approach, including in HIV self-testing programmes.