Four samples each of clear and lightly (thin), moderately, and heavily (thick) meconium-stained amniotic fluid were divided in two portions and submitted twice for assessment to 20 midwives (a total of 320 case assessments). None of the midwives completely agreed with the standard assessment for more than 85 percent of the cases. When disregarding clear samples, for which there was good agreement, each of the midwives classified on average only 35.8 percent of the meconium-stained samples in the same category on each of the four occasions that they were presented to them. Calculation of kappa statistics, which express proportional agreement corrected for chance, indicated that none of the midwives showed very good agreement (kappa > 0.81) with the standard and that fewer than 10 percent showed very good agreement with themselves. The data indicate that grading the severity of meconium staining by visual assessment has such poor accuracy and precision that it cannot provide a valid basis for assigning different care policies to different degrees of meconium staining.