The efficiency of an automotive engine based on a “self-breathing” and “self-humidified” proton exchange membrane fuel cell stack (PEM FC) connected to a dc brushless electrical motor was measured under variable power load conditions. Experiments have been carried out on a small scale 150 W engine model. After determining the fuel cell static polarization curve and the time response to power steps, the system was driven to copy on the test bench a “standard urban load cycle” and its instantaneous efficiencies were measured at an acquisition rate of 5 Hz. The integral system efficiency over the entire urban load cycle, comprising the losses of the unavoidable auxiliary components of the engine, was then calculated. The fuel cell stack was operated mainly in “partial” dead-end mode, with a periodic anode flow channel purging, and one test was carried out in “pure” dead-end mode, with no anode channel purging. An uncertainty analysis of the efficiencies was carried out, taking into account either type A and type B evaluation methods, strengthening the discussion about the outcomes obtained for a system based on this novel simplified FC type. For our small scale engine we measured over the standard urban cycle, on the basis of the H2 high heating value (HHV), a tank-to-wheel integral efficiency of (18.2±0.8)%, when the fuel cell was operated with periodic flow channel purging, and of (21.5±1.3)% in complete dead-end operation mode.