Validity and reliability of upper body push and pull tests to determine one-repetition maximum

Objectives: The purpose of this study was to explore the validity and reliability of three different strength testing approaches to determine one-repetition maximum (1RM) in the bench press and prone bench pull. Methods: Twenty-eight recreationally active subjects (25 ± 2 years, 178 ± 8 cm, 78 ± 9 kg) were assessed for load-velocity (L-V) relationship, 1RM, maximal isometric force (MIF), and maximal repetitions to failure (MRF) in a Smith Machine on three separated sessions. Linear regression was used for L-V relationship, MIF, and MRF to predict 1RM. Level of significance was set to ρ ≤ 0.05. Results: Reliability analyses of the varying 1RM estimations revealed mean differences from 0.6 to -1.3 kg (mainly trivial effects) between test days 1 and 2, intraclass correlation coefficient was > 0.96, and coefficient of variation (CV) was in the range 2.3–8.3% for all tests. Regarding validity, all 1RM predictions exhibited a mean difference ≤ 1.3 kg (trivial), except for the L-V relationship method that underestimated the predicted 1RM by 5 kg (small) compared to the actual bench press 1RM. However, the L-V relationship method showed the least mean absolute errors. CVs were in the range 4.5–13.2%. Standard error of the estimate was in the range 3.2–9.7 kg. Change scores for all tests were significantly correlated with change scores in actual 1RM, except for MIF in the prone bench pull. Smallest deviations in 1RM predictions were observed for the L-V relationship approach. Conclusions: All 1RM prediction methods were highly comparable to the traditional 1RM test. However, given the high variability associated with individual predictions for each method, they cannot be used interchangeably.