Background: The template for the current study was the project conducted by Rosenthal [1973. "An Ecological Study of Free Play in the Nursery School." Doctoral diss., Wayne State University] where she investigated the attraction and holding power of various learning centres during free play in a naturalistic preschool setting. Rosenthal's most significant conclusion was that the characteristics of individual settings were more significant predictors of free play than demographic variables. Purpose: The purpose of this study was to follow Rosenthal's design within a physical education setting to answer three key questions; (i) across lessons, what is the extent and range of engagement across different activity stations, (ii) which activities have a stronger or weaker attractiveness in terms of enticing children to practice, and (iii) which activities have stronger or weaker holding power in terms of their capacity to sustain engagement? Participants and setting: The participants in this study were 12, four-year-old children (10 boys, 2 girls) who attended a daycare centre that serves mostly African American children. The children participated in a biweekly motor programme promoting motor-skill instruction over a ten-week period (~600 minutes). The mastery motivational climate of the programme was based on achievement goal theory and included six to eight motor-skill stations designed to promote locomotor and object-control skills, core balance, spatial awareness, as well as leg and arm strength. Methods: All sessions were videotaped using two wall-mounted cameras located in the top corners of the play area. The total and percentage of time each child spent directly engaged "in" each station was determined, as well as source (self, peer, or teacher) responsible for redirection of activity. Attraction power was measured as the number of different children who visited a station at least one time during a lesson, while holding power was measured as the mean time per focal child entry. Deductive analysis was then conducted by systematically looking for similarities or differences between each activity's attraction/holding power and those elements suggested by Rosenthal that would lead to satiation or prolonged engagement. Findings: The data on initiation and termination showed that 86.1% of station entries were spontaneously instigated by the children themselves, where on average they visited 4.43 different stations per 30-minute session. The most attractive settings were characterized by having elements of novelty, authenticity, and appeared to match the skill of the children. There was a wide range of setting differences in holding power, extending from just a little under a minute's average stay at kicking stations, to involvement lasting four times as long at jumping stations. Holding power was rooted in the extent to which the activity stations were built for success, had the potential for modification, and provided frequent indicators of progress. Conclusions: The major findings from this study would be that inter-setting variation is perhaps a more appropriate target for consideration in the design of learning tasks than is inter-individual variation, because settings are considerably more amenable to teacher control than are individual personalities. Implications for the design of tasks in physical education are included.