We determined inter-rater agreement for the VB-MAPP, an instrument sometimes used in planning educational goals and evaluating intervention effects for young people with autism. A pair of raters independently rated each of 32 children diagnosed with autism. Intraclass correlation coefficients for the total Milestones and Barrier scores were 0.876 and 0.629, respectively, indicating good and moderate reliability. There was variability in reliability in the different domains of the Milestones Assessment, with most indicating moderate reliability, and most of the individual Barriers Assessment domains indicating poor reliability. These are the first data relevant to the reliability of the VB-MAPP, they suggest that further evaluation of its reliability is merited and that a high reliability for individual domains should not be assumed.