This paper presents an empirical study on procedural learning from errors that was conducted within the field of vocational education. It examines whether, and to what extent, procedural learning can benefit more from the detection and written analysis of errors (experimental condition) than from the correct elements (control group). The study involved 136 commercial employee apprentices who were working on a videorecorded, worked-out example of a professional procedure involving customer counselling. The results showed no differences between the conditions with regards to declarative knowledge acquisition during the procedure. Additionally, the analysis of errors was demonstrated to be more effective than the analysis of correct behaviours in terms of (1) representing a much more detailed and correct procedure and (2) developing anticipatory knowledge about possible errors to be avoided. The results highlighted the value of the inductive instructional strategy (from the analysis of a single worked-out example to the learning of the general procedure) and opened the field up for future applications of a similar instructional approach combining roleplaying and video analysis of errors in vocational education to develop communication skills.