The effects of latency on the accuracy of data recorded by three special education teachers were examined in this study. Teachers recorded data on the target behaviors of three students with varying disabilities. The accuracy of data recorded was assessed during three time periods: immediately after the target behavior occurred, at the end of the school day, and the following school day. A multielement design was used to evaluate data accuracy. Results were interpreted to confirm that data recorded immediately after a behavior occurred were more accurate and reliable than data documented at the end of the school day or the start of the following school day. In addition, data recorded by each teacher had a mean agreement of 97% or above for the time period immediately after a student's behavior occurred. Furthermore, each teacher reported that it was beneficial to record data immediately after the target behavior occurred. Implications and future research directions are provided. (Contains 3 figures and 3 tables.)