We evaluated the reliability (i.e. reproducibility and variability) in measuring regional gastric mucosal blood flow by laser Doppler flowmetry in a total of 32 subjects (12 normal subjects and 20 patients with cirrhosis). The measurements were performed in quadruplicate within 10 min by observer A (A1, A2) and B (B1, B2) using blind techniques. The intraclass correlation coefficient (Ri) was used to assess observer agreement. Overall, the level of intra- (Ri, 0.87 for A1 vs. A2 and 0.87 for B1 vs. B2) and interobserver agreement (0.86 for A vs. B) was excellent. In contrast, there was a higher observer variability assessed by coefficient of variation. Overall, the mean value of coefficient of variation was 12% (max. 35%) for A1 vs. A2, 12% (max. 35%) for B1 vs. B2, and 13% (max. 33%) for A vs. B. These results suggest that laser Doppler flowmetry, although the reproducibility may be clinically acceptable, seems to still be a subjective method for quantitation of gastric mucosal blood flow in humans. Therefore, studies using this method can only be acceptable when the changes that investigators intend to measure are significantly higher than the variability.