Central venous catheterization (CVC) is a complex but commonly performed procedure. How best to teach this complex skill has not been clearly delineated. We conducted a randomized trial of the effects of two types of teaching of CVC on skill acquisition and retention. We randomly assigned novice internal medicine residents to learning CVC in-part or in-whole. The part-group was taught the first part of the procedure, followed by practice, followed by being taught the second and final portion of the procedure, and followed by practice. The whole-group was taught the procedure in its entirety, followed by practice. Teaching and practice time for both groups was otherwise held constant. Performances were assessed at baseline, post-training, and at 1 month. The primary outcome was skill retention at 1-month, rated by using a global rating scale and a 22-item checklist, and defined as the score increase between 1-month and baseline. Skill acquisition is defined as the score increase post-training and baseline. Raters were blinded to the participants' identity, group assignment, and time point. Participants in the part-task group outperformed the whole-task group in skill acquisition (2.2 ± 0.8 vs 1.3 ± 1.0; g = 1.01; p = 0.04) and in skill retention (1.5 ± 0.7 vs 0.5 ± 0.8; g = 1.39; p = 0.006) using the global rating scale. Scores rated by the checklist were not significantly different (52.0 ± 25.3 vs 43.5 ± 23.4; g = 0.33; p = 0.47 for skill acquisition; and 48.5 ± 34.9 vs 41.1 ± 20.4; g = 0.35; p = 0.44 for skill retention). For teaching ultrasound-guided CVC to novice learners, teaching in part is preferable than teaching in whole.