Abstract: Background. There are many types of tests of injury and injury protection in which some proxy for injury is observed in controlled test conditions. This paper considers the implications of a test result for the consequences averaged over the variety of different conditions that occur in the real world. Method. Theoretical analysis of what a test result implies, linked to two examples of applications. The first example is incapacitation from missiles such as bullets or fragments, where averaging is over the body surface. The second is head injury in road accidents, where averaging is over speeds of impact. Results. An equation for average consequences is obtained, based on (a) generalising what happens under one set of conditions to different conditions, (b) transforming what happens in the test to something (a value or a cost) that it is meaningful to average, and (c) using probabilities of different conditions to compute the average consequences. This equation both clarifies earlier work and demonstrates common features of all testing protocols if it is desired to use one set of conditions to represent a great variety of real-world sets of conditions.