Abstract The object of this paper is to illustrate the type of laboratory test data which a turbine manufacturer must have to predict intelligently the runaway speed of a Kaplan turbine. The significant differences between the “on-cam” and “off-cam” runaway speeds are explained and data on actual runaway-speed tests conducted on a large Kaplan turbine installation are given. Attention is called to the possibility of reducing the runaway speed by restricting the minimum angle to which the runner blades can be pitched, and information is presented that will aid the purchaser of a Kaplan turbine to decide what runaway speed should be provided for in the generator.