The purpose of this article is a close examination of the evidence on error rate on classroom tasks. The finding of a negative relationship between error rate, especially rate of oral reading errors, and gains in reading achievement is generally interpreted to mean that a low rate of error leads to growth in reading. However, we argue that this relationship may be an epiphenomenon: (1) error rate on classroom tasks is a good measure of children's level of reading development, (2) standardized tests and other one-shot assessment instruments are always imperfect measures of reading level, (3) error rate correlates negatively with end-of-year achievement because it provides additional and more reliable information, beyond that contained in previous test scores, about children's reading ability. Findings from a microanalytic study of third-grade reading lessons confirmed that oral reading errors can have a positive influence on children's comprehension. Errors facilitated comprehension of non-turn-takers when task norms emphasized accurate oral reading but not when norms emphasized story understanding. According to our theory, an oral reading error followed by feedback fits the pattern of tension followed by resolution shown by other research to improve learning and memory. When the task is accurate oral reading, a failure to read fluently produces tension, which increases attention and instigates deeper processing or a greater "effort after meaning."