Ultrasound simulation allows students to virtually explore internal anatomy by producing accurate, moving, color, three-dimensional rendered slices from any angle or approach leaving the organs and their relationships intact without requirement for consumables. The aim was to determine the feasibility and efficacy of self-directed learning of cardiac anatomy with an ultrasound simulator compared to cadavers and plastic models. After a single cardiac anatomy lecture, fifty university anatomy students participated in a three-hour supervised self-directed learning exposure in groups of five, randomized to an ultrasound simulator or human cadaveric specimens and plastic models. Pre- and post-tests were conducted using pictorial and non-pictorial multiple-choice questions (MCQs). Simulator students completed a survey on their experience. Four simulator and seven cadaver group students did not attend after randomization. Simulator use in groups of five students was feasible and feedback from participants was very positive. Baseline test scores were similar (P = 0.9) between groups. After the learning intervention, there was no difference between groups in change in total test score (P = 0.37), whether they were pictorial (P = 0.6) or non-pictorial (P = 0.21). In both groups there was an increase in total test scores (simulator +19.8 ±12.4%% and cadaver: +16.4% ± 10.2, P < 0.0001), pictorial question scores (+22.9 ±18.0%, 19.7 ±19.3%, P < 0.001) and non-pictorial question scores (+16.7 ±18.2%, +13 ±15.4%, P = 0.002). The ultrasound simulator appears equivalent to human cadaveric prosections for learning cardiac anatomy.