BACKGROUND: CD4(+) T-cell enumeration (CD4 count) is used as a criterion to initiate antiretroviral therapy (ART) in HIV patients and to monitor treatment efficacy. However, simple, affordable, and reliable point-of-care (POC) instruments adapted to resource-limited settings are still lacking. The PIMA CD4 analyzer is a new POC instrument for CD4 counting that uses disposable cartridges and a battery-powered analyzer. METHODS: Whole blood samples were taken by venipuncture or by finger prick from 300 subjects, including HIV-infected patients and HIV (-) controls. CD4 counts were measured by PIMA (using venous or capillary blood) and by FACSCount (using venous blood) considered as the reference. RESULTS: Similar CD4 counts were obtained by PIMA and FACSCount using either HIV+ venous blood or HIV+ finger-prick blood samples. However, with a concordance coefficient of 0.88 and a Pearson correlation of 0.89, finger-prick blood performed not as good as venous blood (0.97 and 0.98, respectively). For a clinical decision to start ART at 200 CD4 cells per microliter, sensitivity of PIMA was 90%/91% and specificity 98%/96% for venous/finger-prick blood, respectively, and for a treatment threshold of 350 CD4 cells per microliter, the sensitivity was 98%/91% and the specificity was 79%/80% for venous/finger-prick blood, respectively. Repeatability (precision) on venous blood resulted in a coefficient of variation of 4%. Using finger-prick blood, the average instrument error frequency resulting in aborted analyses was 14%. CONCLUSIONS: PIMA is a good POC instrument for screening adult HIV-infected patients in resource-limited settings for treatment eligibility. Its performance on finger-prick blood is not as good as on venous blood. Adequate training for correct use of finger-prick blood samples is mandatory.