Development of reliable, easy-to-use, rapid diagnostic tests (RDTs) to detect glucose-6-phosphate dehydrogenase (G6PD) deficiency at point of care is essential to deploying primaquine therapies as part of malaria elimination strategies. We assessed a kit under research and development called CareStart™ G6PD deficiency screening test (Access Bio, New Jersey, USA) by comparing its performance to quantitative G6PD enzyme activity using a standardized spectrophotometric method ('gold standard'). Blood samples (n = 903) were collected from Cambodian adults living in Pailin province, western Cambodia. G6PD enzyme activities ranged from 0 to 20.5 U/g Hb (median 12.0 U/g Hg). Based on a normal haemoglobin concentration and wild-type G6PD gene, the normal values of G6PD enzymatic activity for this population was 3.6 to 20.5 U/g Hg (95(th) percentiles from 5.5 to 17.2 U/g Hg). Ninety-seven subjects (10.7%) had <3.6 U/g Hg and were classified as G6PD deficient. Prevalence of deficiency was 15.0% (64/425) among men and 6.9% (33/478) among women. Genotype was analyzed in 66 G6PD-deficient subjects and 63 of these exhibited findings consistent with Viangchang genotype. The sensitivity and specificity of the CareStart™ G6PD deficiency screening test was 0.68 and 1.0, respectively. Its detection threshold was <2.7 U/g Hg, well within the range of moderate and severe enzyme deficiencies. Thirteen subjects (1.4%, 12 males and 1 female) with G6PD enzyme activities <2 U/g Hg were falsely classified as "normal" by RDT. This experimental RDT test here evaluated outside of the laboratory for the first time shows real promise, but safe application of it will require lower rates of falsely "normal" results.