We have developed a device to measure lightning-channel propagation velocities. It consists of eight solid state silicon photodetectors mounted behind precision horizontal slits in the focal plane of a 50-mm lens on a 35-mm camera body. Each detector has a 0.1° vertical field of view that is separated from adjacent detector slits by 2.8°. The horizontal field-of-view for each detector is 41° and the total vertical field of view for the device is 21°. The signal from each detector is amplified by a circuit with a 10%–90% rise time of 0.6 μs and an equivalent decay time of 400 μs. The eight Photodetector pulses, IRIG-B time, and slow and fast electric field change waveforms are recorded on a 14-track analog tape recorder with an upper frequency response Of 1.0 MHz and a maximum dynamic interchannel timing error of 0.6 μs. To provide images of lightning geometry and permit time-to-thunder measurements, color video and sound are recorded with a standard VHS video cassette recorder. The return stroke velocity (RSV) device, video camera, and microphone are installed and coaxially aimed in an environmental enclosure on a remotely controlled pan-tilt unit atop our mobile laboratory, permitting the recording of lightning signals at remote sites and while mobile. To evaluate the performance of the RSV device, we have analyzed 12 natural return strokes from Alabama, Florida, and Oklahoma and 4 return strokes triggered at the Kennedy Space Center, Florida. The velocities we determined vary from 1.2 to 2.5×108 m s−1, with an average of 1.7×8 m s−1±0.8 × 8 m s−1. From comparisons of our results to those of a streaking camera, we find no significant differences between the velocities obtained from the same strokes with the two systems. We also find no differences between the characteristics of the pulses or the velocities calculated from them while the RSV device is moving or stationary.