BACKGROUND: Blood glucose testing (BGT) at the forearm minimizes the pain experienced during sampling of capillary blood. We compared the BGT results for forearm sampling with those for standard finger skin puncture and venous serum to evaluate the clinical validity of forearm BGT. METHODS: BGT was performed on the finger (G(F)) and forearm (G(A)) with a portable glucometer in 555 subjects, including 61 diabetic patients, under fasting conditions. BGT with venous serum (G(V)) was followed within an hour in 514 subjects. Simple linear regression, intraclass correlation, and Passing-Bablok regression analyses were performed using the G(A)-G(F) and G(A)-G(V) data. RESULTS: G(A) showed an excellent linear relationship with both G(F) and G(V) with a Pearson correlation coefficient (r) of 0.97 (P<0.0001) in the patient group, which was similar to the findings in the normal group except for the lower r values. The mean bias between G(A) and G(F) and between G(A) and G(V) were within +/- 10 mg/dL in both groups. The intraclass correlation coefficients were slightly smaller than the corresponding r values, but they showed the same tendency in both groups. In the Passing-Bablok analyses, the 95% confidence intervals of the slope and intercept parameters were <+/-20% of unity and <+/-20 mg/dL, respectively, which were within the acceptable ranges. All 3 statistical analyses supported the satisfactory agreement of G(A) with G(F) or G(V). CONCLUSIONS: BGT at the forearm was highly consistent with the standard BGT, thereby confirming its applicability in clinical practice for self-testing under steady fasting conditions.