Athletes' social interactions represent an important context through which positive development can be fostered (Fraser-Thomas & Côté, 2009). However, the existing literature has relied primarily on the use of retrospective or self-report measures that may not fully capture the complexity of athletes' behaviours (Meredith, Dicks, Noel, & Wagstaff, 2018). As such, there is a need to use observational techniques to explore what athlete behaviours look like in naturalistic settings (Smith, 2003). Thus, the purpose of this case study was to use observational methods to understand athletes' engagement in activities and social interactions within the context of an English football academy. Ten athletes in an under-11 academy were observed over seven practices. Practices were purposefully sampled to represent a diverse range of contexts (e.g., practice activities, scrimmages, and team meetings). Athletes' behaviours were coded using an adapted version of the Athlete Behaviour Coding System, which assesses behaviours according to their content, target, quality, and the context in which they occur (Vierimaa & Côté, 2016). Findings indicated that athletes most commonly exhibited communicative behaviours (M = 359.54 counts/hour, SD = 182.40 counts/hour), followed by engaged (M = 236.31 counts/hour, SD = 223.56 counts/hour) and non-cooperative behaviours (M = 1.38 counts/hour, SD = 5.12 counts/hour). It was also found that athletes touched the ball more during practice activities (M = 99.17 counts/hour, SD = 54.97 counts/hour) than during scrimmages (M = 90 counts/hour, SD = 60 counts/hour). The theoretical and practical implications of this research for coaches and sport practitioners will be discussed.