Several recent studies have shown that children with autism spectrum disorder (ASD) are less physically active than their typically developing peers and are thus at risk of health problems associated with inactivity (e.g., obesity or diabetes). There is a need to examine the effectiveness of interventions such as swimming that are aimed at increasing physical activity, using approaches that are ecologically valid--that is, that are designed to address real-world problems in natural settings with typical intervention agents. This community-based study used a modified non-concurrent multiple baseline design to assess the impact of behavioral skills training to train recreation center staff in the use of discrete trials and visual activity schedules to teach swimming skills to children with ASD in public pools. Following completion of the training, five of six instructors showed evidence of instructional skill acquisition; the sixth instructor showed improvement during baseline, so the effect of the intervention could not be assessed. In addition, an experimental effect for increased compliance was evident for seven of the eight children, and all eight acquired new swimming skills. Social/ecological validity ratings by instructors and parents were uniformly high. The results and limitations are discussed, along with suggestions for future research.