There is considerable demand for information on the effectiveness of various resistance exercises for improving physical performance, and on how exercise programs must match functional activities to produce the greatest performance gains (training specificity). Evidence supports exercise-type specificity; the greatest training effects occur when the same exercise type is used for both testing and training. Range-of-motion (ROM) specificity is supported; strength improvements are greatest at the exercised joint angles, with enough carryover to strengthen ROMs precluded from direct training due to injury. Velocity specificity is supported; strength gains are consistently greatest at the training velocity, with some carryover. Some studies have produced a training effect only for velocities at and below the training velocity while others have produced effects around the training velocity. The little, mainly isokinetic, evidence comparing different exercise velocities for improving functional performance suggests that faster exercise best improves fast athletic movements. Yet isometric exercise can improve actions like the vertical jump, which begin slowly. The rate of force application may be more important in training than actual movement speed. More research is needed into the specificity and efficacy of resistance exercise. Test populations should include both males and females of various ages and rehabilitation patients.