ABSTRACT Background: Coach observation studies have been a mainstay of coaching science research for decades, with a major focus on the use of instruction and feedback by coaches. However, relatively few studies have investigated feedback provided in a live competition setting, with a majority focussing on the training environment. As such, little is known about the way that feedback changes as a result of the intensity of a competitive setting. Additionally, feedback variables of interest are featured inconsistently between studies. Purpose: The purpose of the current study was to investigate the quantity and nature of feedback from coach to player in an Australian rules football context, specifically the elite-level Australian Football League (AFL) competition. An entire season's worth of coach-player feedback was recorded and coded against a wide range of feedback variables most commonly found in the literature. A secondary aim of the study was to examine changes in feedback against various game metrics (winning/losing quarter, margin). Method: Five coaches from one AFL team, all with several years of experience at the elite level, were recorded across 21 regular-season games from the 2018 season. Feedback quantity, along with several feedback variables of interest (valence, autonomy-support, descriptive/prescriptive, individual/group, etc.), were coded during each quarter of play. Game information (game outcome, quarter outcome, margin at the start of the quarter, point differential for the quarter) was also recorded. Results: Across 84 quarters of football, 1301 coach-player feedback instances were observed. Mean values across all quarters revealed more negative (20%) than positive (13%) feedback, more prescriptive (81%) than descriptive (19%) feedback, more controlling (58%) than autonomy-supportive (8%), and more individual (70%) than group (30%) feedback was provided. Task-related feedback (60%) was more common than process-related (37%), self-regulation feedback (3%) or self-level feedback i.e. praise (2%). More positive feedback was provided in winning quarters than in losing quarters, while more controlling feedback was present during losing quarters than during winning quarters. Coaches provided significantly more feedback overall when the game was in the balance than quarters where the team was 95% or more likely to win or lose. Conclusions: The paper presents novel findings about feedback use in an elite competitive team sport setting. Areas where the use of feedback does not align with current theory and evidence are highlighted, particularly regarding feedback valence and autonomy support. Feedback variables more commonly associated with improved learning and performance in the literature appear to generally be provided more frequently during winning quarters, but a causal relationship is yet to be determined.