Legal mandates and best practice recommendations for the education of students with autism spectrum disorders (ASD) emphasize the importance of systematic, ongoing observational data collection in order to monitor progress and demonstrate accountability. The absence of such documentation in decision-making on instructional objectives indicates a weakness in bridging the research-to-practice gap in special education. Utilizing a multiple baseline design across participants, the current study evaluated the effects of a prototypical teacher training program (i.e., workshop, checklist, in-classroom training with feedback, and maintenance with a thinned schedule of feedback) on the frequency of data collection on core deficits of ASD and the use of data-based decision-making. Results indicate increases in daily mean frequency of data collection following intervention. Maintenance and generalization indicates variable responding across participants. Effect size (Cohen's d) indicates a large, clinically significant effect of the training program. Results are discussed in relation to training models, maintenance, and future research.