ObjectiveKnee arthrocentesis is frequently performed as a diagnostic or therapeutic procedure. Although considered a key competency for medical doctors, most students never execute the procedure during their training. We aimed to assess technical and nontechnical skills for knee arthrocentesis through high-fidelity hybrid simulation.DesignTrainees received educational documentation prior to training. Trainees took a medical history and obtained informed consent from a patient-actor, then encountered a simulated knee to execute the procedure. We adapted a direct observation scale to assess technical and nontechnical skill performance. Personalized feedback was received after each session. Performance among trainees (learning curves) and between trainees and experts was compared using a mixed-effects model.SettingPontifical Catholic University of Chile Simulation Center.ParticipantsMedical students and general physicians were recruited for training; orthopedic surgeons were recruited as experts.ResultsTrainees significantly improved from the first session to the second and third. The third session was the learning curve plateau. Performance obtained in the third and fourth sessions were similar to expert performance. The assessment tool evaluated technical and nontechnical skills with high internal consistency and showed high interobserver reliability.DiscussionLearning curve analysis showed that high-fidelity simulation allowed trainees to become proficient in technical and nontechnical skills required to perform a safe knee arthrocentesis.Level of evidenceLevel II (prospective cohort study).