Abstract: The quantification of CD4 T-lymphocytes, an important marker of HIV disease progression, signals initiation of or regulation of antiretroviral therapy. The study assessed the accuracy of CD4 tests results generated at five public health laboratories in Kisumu County- Kenya. Samples from HIV positive patients that had been tested for CD4 at five high volume (>1000 tests/month), high proximity (within 100km radius of Kisumu county) laboratories were retested at the reference laboratory. Results were entered into a Microsoft excel sheet and analyzed to determine the percentage coefficient of variation (CV), the mean difference and the concordance correlation coefficient between results generated at the public health laboratories and the local reference laboratory. A coefficient of variation of ≤ 13 %, a mean difference of ±50 cells/µl and a concordance correlation coefficient of ≥ 0.90 were considered to be within acceptable limits. Overall, repeat test results had a mean CV of 10.5% (Range: 8.9%-11.4%) and a mean difference of -15 cells/µl (Range: -3.3; -47.2). Laboratory A had the least coefficient of variation of 8.9%, the least bias; -3.33 cells/ µl (-119.59; -112.93) and the highest concordance correlation coefficient of (0.98) hence most accurate of all. The CD4 results at all laboratories were comparable to the reference laboratory results. The CD4 Proficiency Testing (PT) program is a feasible alternative to formal costly External Quality Assessment (EQA) programs for CD4 testing. Scale up of repeat testing PT could encourage continuous quality improvement in CD4 testing among laboratories in Kenya.