Introduction: Pericardiocentesis is a potentially life-saving procedure. We compared two low-cost models—an agar-based model and a novel model, Centesys—in terms of ultrasound image quality and realism, effectiveness of the model, and learners’ confidence and satisfaction after training. Methods: In this pilot randomised 2x2 crossover trial stratified by physician seniority, participants were assigned to undergo pericardiocentesis training either with the agar-based or Centesys model first, followed by the other model. Participants were asked to rate their confidence in performing ultrasound-guided pericardiocentesis, clarity and realism of cardiac structures on ultrasound imaging, and satisfaction on a 7-point Likert scale before and after training with each model. Results: Twenty participants with median postgraduate year of 4 (interquartile range [IQR] 3.75–6) years were recruited. Pre-training, participants rated themselves a median score of 2.5 (IQR 2–4) for level of confidence in performing pericardiocentesis, which improved to 5 (IQR 4–6) post-training with Centesys (P=0.007). Centesys was recognised to be more realistic in simulating cardiac anatomy on ultrasound (median 5 [IQR 4–5] versus 3.5 [IQR 3–4], P=0.002) than the agar-based model. There was greater satisfaction with Centesys (median 5 [IQR 5–6] versus 4 [IQR 3.75–4], P
