Economic pressures on universities and the emergence of new technologies have spurred the creation of new systems for delivering engineering laboratories in education, in particular simulations and remote-access laboratory systems. Advocates of simulation argue that physical labs needlessly consume university space and students' time. However, proponents of hands-on laboratories argue that student engineers should be exposed to real environments. Remote laboratories have appeared as a third option. These laboratories are similar to simulation techniques in that they require minimal space and time, because the experiments can be rapidly configured and run over the Internet. But unlike simulations, they provide real data. Studying the relative effectiveness of these modes of delivering student laboratories is complex, for the underlying technology of the laboratory is just one of many possible factors that could influence effectiveness. For example, the interface to the equipment may be of importance, as might the discussions students have among themselves. This paper presents a model for testing the relative effectiveness of engineering laboratories in education that takes account of these and other factors. The results are presented for an assessment study comparing versions of remote labs versus hands-on labs in a junior-level mechanical engineering course on machine dynamics and mechanisms. The results suggest that students learned lab content information equally well from both types of laboratories, and that they have a realistic understanding and appreciation of the practical advantages of remote laboratories.