The development of sport-specific dynamometers is an important step towards ecological validity in analysing athlete performance. Design limitations in previous punch-measuring devices have resulted in values which may not or cannot fully reflect the force and multidirectional components in a punch. In developing this boxing dynamometer, a triaxial force measurement system and a boxing manikin interface were combined. The repeatability and accuracy of the dynamomoter were assessed using simulated straight punches. Discrimination efficacy was assessed by comparison of the maximal punching force of seven elite, eight intermediate and eight novice boxers during simulated boxing, throwing straight punches. For the elite, intermediate and novice groups, respectively, the maximal straight punching forces (mean +/- s(mean)) were 4800 +/- 227 N, 3722 +/- 133 N and 2381 +/- 116 N for the rear hand, and 2847 +/- 225 N, 2283 +/- 126 N and 1604 +/- 97 N for the lead hand. For all groups, maximal forces were larger for the rear than the lead hand (P < 0.001). Maximal punching force was greater in the elite than the intermediate group, and greater in the intermediate than the novice group (P < 0.05). The boxing dynamometer discriminated effectively between punching performance at three standards of performance and between the punching force of the rear and lead hands.