Working with middle-school students, we replicated and extended research on oral reading fluency (ORF) assessments and prompting students to read faster. Altering ORF administration procedures by instructing students to read fast caused statistically significant increases in their words correct per minute (WCPM) and errors, which was moderated by reading skills. Specifically, prompts to read fast caused statistically significant larger increases in WCPM and smaller increases in errors among the stronger readers, relative to the weaker readers. In Study 2, this moderating effect was not present. During Study 2, participants instructed to read fast received subsequent instructions to do their best reading or to do their best reading because they would answer questions. These instructions worked to slow participants' reading speed. Implications for prompting faster reading, future research into causal variables for the skill level-by-instructions interaction revealed in Study 1, and the need for comprehension measures in ORF assessments is discussed.