Contribution: This article analyzes the learning effectiveness of a virtual educational escape room for teaching software engineering and compares this activity with traditional teaching through a randomized controlled trial. Background: Educational escape rooms have been used across a wide variety of disciplines at all levels of education and they are becoming increasingly popular among teachers. Nevertheless, there is a clear general need for more robust empirical evidence on the learning effectiveness of these novel activities and, particularly, on their application in software engineering education. Research Questions: Is game-based learning using educational escape rooms more effective than traditional lectures for teaching software engineering? What are the perceptions of software engineering students toward game-based learning using educational escape rooms? Methodology: The study presented in this article is a randomized controlled trial with a pre- and post-test design that was completed by a total of 326 software engineering students. The 164 students belonging to the experimental group learned software modeling by playing an educational escape room whereas the 162 students belonging to the control group learned the same subject matter through a traditional lecture. Findings: The results of the randomized controlled trial show that the students who learned software modeling through the educational escape room had very positive perceptions toward this activity, significantly increased their knowledge, and outperformed those students who learned through a traditional lecture in terms of knowledge acquisition.