While studies have evaluated the accuracy of adult patients and health personnel in reading various glucose oxidase impregnated strips to estimate blood glucose, there are no studies exclusively evaluating the accuracy of children with diabetes reading their own strips as compared to a staff member, and meter to meter variability in reading these strips. We evaluated the accuracy of reading chemstrip bG by children at a summer camp. The children's visual readings of their own strips were compared to the visual reading of a single staff member. A total of 356 Chemstrip bG's were visually read by diabetic children and a single trained staff member at a summer camp for diabetics. The strips were then analyzed by two Accu-Chek bG meters. Intermachine variability was found to be negligible over the entire bG range. For the purposes of this study, we define accurate visual readings as those within +/- 15 percent of the meter reading of a given strip. At low bG values (40-79 mg/dl), accuracy by children and staff is low, with underestimating occurring in 39 percent of staff readings and 57 percent of children's readings. At intermediate bG values (120-239 mg/dl) readings are more accurate, especially when read by the staff, with misreadings occurring in only 16-19 percent of the strips. At high bG values (240-399 mg/dl), accuracy by children is decreased, with underestimation 500 percent more often than staff. We conclude that children are less accurate at reading Chemstrip bG than a trained staff member (51% versus 33% misreading), especially at the upper and lower ranges of bG values when visual readings are least accurate, and the need for therapeutic intervention is the greatest.