In the quest to maximize average propulsive stroke impulses over 2000-m racing, testing and training of various strength parameters have been incorporated into the physical conditioning plans of rowers. Thus, the purpose of this review was 2-fold: to identify strength tests that were reliable and valid correlates (predictors) of rowing performance; and, to establish the benefits gained when strength training was integrated into the physical preparation plans of rowers. The reliability of maximal strength and power tests involving leg extension (e.g. leg pressing) and arm pulling (e.g. prone bench pull) was high (intra-class correlations 0.82-0.99), revealing that elite rowers were significantly stronger than their less competitive peers. The greater strength of elite rowers was in part attributed to the correlation between strength and greater lean body mass (r = 0.57-0.63). Dynamic lower body strength tests that determined the maximal external load for a one-repetition maximum (1RM) leg press (kg), isokinetic leg extension peak force (N) or leg press peak power (W) proved to be moderately to strongly associated with 2000-m ergometer times (r = -0.54 to -0.68; p < 0.05). Repetition tests that assess muscular or strength endurance by quantifying the number of repetitions accrued at a fixed percentage of the strength maximum (e.g. 50-70% 1RM leg press) or set absolute load (e.g. 40 kg prone bench pulls) were less reliable and more time consuming when compared with briefer maximal strength tests. Only leg press repetition tests were correlated with 2000-m ergometer times (e.g. r = -0.67; p < 0.05). However, these tests differentiate training experience and muscle morphology, in that those individuals with greater training experience and/or proportions of slow twitch fibres performed more repetitions. Muscle balance ratios derived from strength data (e.g. hamstring-quadriceps ratio <45% or knee extensor-elbow flexor ratio around 4.2 ± 0.22 to 1) appeared useful in the pathological assessment of low back pain or rib injury history associated with rowing. While strength partially explained variances in 2000-m ergometer performance, concurrent endurance training may be counterproductive to strength development over the shorter term (i.e. <12 weeks). Therefore, prioritization of strength training within the sequence of training units should be considered, particularly over the non-competition phase (e.g. 2-6 sets × 4-12 repetitions, three sessions a week). Maximal strength was sustained when infrequent (e.g. one or two sessions a week) but intense (e.g. 73-79% of maximum) strength training units were scheduled; however, it was unclear whether training adaptations should emphasize maximal strength, endurance or power in order to enhance performance during the competition phase. Additionally, specific on-water strength training practices such as towing ropes had not been reported. Further research should examine the on-water benefits associated with various strength training protocols, in the context of the training phase, weight division, experience and level of rower, if limitations to the reliability and precision of performance data (e.g. 2000-m time or rank) can be controlled. In conclusion, while positive ergometer time-trial benefits of clinical and practical significance were reported with strength training, a lack of statistical significance was noted, primarily due to an absence of quality long-term controlled experimental research designs.