The ability to optimise muscular power output is considered fundamental to successful performance of many athletic and sporting activities. Consequently, a great deal of research has investigated methods to improve power output and its transference to athletic performance. One issue that makes comparisons between studies difficult is the different modes of dynamometry (isometric, isokinetic and isoinertial) used to measure strength and power. However, it is recognised that isokinetic and isometric assessment bear little resemblance to the accelerative/decelerative motion implicit in limb movement during resistance training and sporting performance. Furthermore, most people who train to increase power would have limited or no access to isometric and/or isokinetic dynamometry. It is for these reasons and for the sake of brevity that the findings of isoinertial (constant gravitational load) research will provide the focus of much of the discussion in this review. One variable that is considered important in increasing power and performance in explosive tasks such as running and jumping is the training load that maximises the mechanical power output (Pmax) of muscle. However, there are discrepancies in the research as to which load maximises power output during various resistance exercises and whether training at Pmax improves functional performance is debatable. There is also some evidence suggesting that Pmax is affected by the training status of the individuals; however, other strength variables could quite possibly be of greater importance for improving functional performance. If Pmax is found to be important in improving athletic performance, then each individual's Pmax needs to be determined and they then train at this load. The predilection of research to train all subjects at one load (e.g. 30% one repetition maximum [1RM]) is fundamentally flawed due to inter-individual Pmax differences, which may be ascribed to factors such as training status (strength level) and the exercise (muscle groups) used. Pmax needs to be constantly monitored and adjusted as research suggests that it is transient. In terms of training studies, experienced subjects should be used, volume equated and the outcome measures clearly defined and measured (i.e. mean power and/or peak power). Sport scientists are urged to formulate research designs that result in meaningful and practical information that assists coaches and strength and conditioning practitioners in the development of their athletes.