CD4 T-cell counts are widely used to assess treatment eligibility and to follow-up HIV-infected patients. The World Health Organization prequalification of in vitro diagnostics program conducted a performance evaluation of the FACSPresto (BD Biosciences), a new point-of-care instrument to measure absolute CD4-T cell (CD4) counts and percentages in venous and capillary blood samples from HIV-infected patients.Patients were recruited in Belgium (200 patients) and in Tanzania (247 patients). Venous blood samples were analyzed in two nearby reference laboratories. In addition, nurses/technicians collected a capillary blood sample by finger prick directly into a FACSPresto CD4 cartridge. Assay precision was assessed on fresh blood and on external quality control samples. Trueness (bias) was assessed by comparing results from FACSPresto with the reference (single-platform FACSCalibur). Clinical misclassification was measured at 200, 350 and 500 cells/μL thresholds.Intra-assay precision was < 6%, and inter-assay < 8%. CD4 results from FACSPresto and reference method resulted in regression slopes of 0.99-1.11 using either venous or capillary blood. Correlation was better for venous than for capillary blood (minimum 0.97 vs 0.93 respectively). Capillary blood resulted in a larger bias than venous blood, with 24 and 83 cells/μL for absolute CD4 counts on capillary blood in Antwerp and Dar es Salaam respectively, vs 12 and 41 cells/μL on venous blood. Bias on CD4% was < 1% on both venous and capillary blood, and was proportionally better than for absolute CD4 counts. Clinical misclassification was in line with the average overestimation, showing a very good specificity, but sensitivity around 70-90%. The rejection rate was 11% on first reading, leading to 6% of all samples without final result after a second reading.The FACSPresto performed very well on venous blood samples, and well on capillary blood samples.