To test the hypothesis that low-cost homemade models may be used to acquire the basic skills for ultrasound-guided central vein puncture. Training study. University transplantation department. Training was performed using three different homemade models (A, B, and C). Segments of a common rubber tourniquet (V1) and Silastic tube (V2) were used to simulate vessels within agar-based models. Overall cost for each model was less than 5 euro (US$7). For each test (test I, A-V1; II, A-V2; III, B-V1; IV, C-V2), the number of punctures and attempts needed to locate the needle inside the lumen were recorded. Each test was considered completed when participants punctured the vessels at the first attempt for three consecutive times. In test I, the mean number of punctures and attempts were 3.85 +/- 1.26 and 4.95 +/- 3.05; in test II, 4.60 +/- 1.14 and 6.30 +/- 2.51; in test III, 4.80 +/- 1.06 and 4.65 +/- 2.21; and in test IV, 4.45 +/- 1.23 and 6.05 +/- 2.92, respectively. For each test, no statistical difference was found by comparison of number of punctures and attempts for anesthesiologists versus nonanesthesiologists, men versus women, or previous experience versus no experience with central vein cannulation (CVC). Video game users obtained better results than did nonusers in test I (punctures, P = 0.033; attempts, P = 0.038), test II (punctures, P = 0.052; attempts, P = 0.011), and test IV (punctures, P = 0.001; attempts, P = 0.003). A posttraining questionnaire showed favorable opinions about the clarity of the instructions, aptness of the models, and adequacy of the training. In our operative unit, the use of ultrasound guidance for CVC increased from 2% to 23% in the first month after training. Low-cost homemade models are useful in acquiring basic coordination skills for ultrasound-guided CVC.