To investigate the effectiveness of brief bedside "booster" cardiopulmonary resuscitation (CPR) training to improve CPR guideline compliance of hospital-based pediatric providers. Prospective, randomized trial. General pediatric wards at Children's Hospital of Philadelphia. Sixty-nine Basic Life Support-certified hospital-based providers. CPR recording/feedback defibrillators were used to evaluate CPR quality during simulated pediatric arrest. After a 60-sec pretraining CPR evaluation, subjects were randomly assigned to one of three instructional/feedback methods to be used during CPR booster training sessions. All sessions (training/CPR manikin practice) were of equal duration (2 mins) and differed only in the method of corrective feedback given to participants during the session. The study arms were as follows: 1) instructor-only training; 2) automated defibrillator feedback only; and 3) instructor training combined with automated feedback. Before instruction, 57% of the care providers performed compressions within guideline rate recommendations (rate >90 min(-1) and <120 min(-1)); 71% met minimum depth targets (depth, >38 mm); and 36% met overall CPR compliance (rate and depth within targets). After instruction, guideline compliance improved (instructor-only training: rate 52% to 87% [p .01], and overall CPR compliance, 43% to 78% [p < .02]; automated feedback only: rate, 70% to 96% [p = .02], depth, 61% to 100% [p < .01], and overall CPR compliance, 35% to 96% [p < .01]; and instructor training combined with automated feedback: rate 48% to 100% [p < .01], depth, 78% to 100% [p < .02], and overall CPR compliance, 30% to 100% [p < .01]). Before booster CPR instruction, most certified Pediatric Basic Life Support providers did not perform guideline-compliant CPR. After a brief bedside training, CPR quality improved irrespective of training content (instructor vs. automated feedback). Future studies should investigate bedside training to improve CPR quality during actual pediatric cardiac arrests.