Background: The maximal one-repetition test (1-RM) is widely used in scientific research; however, there are conflicting results regarding its reproducibility in elderly populations. The present study aimed to analyze the reproducibility of the test both before and after a 12-week training period by using the bench press and leg press 45° 1-RM tests in the elderly, taking into consideration the training experience and strength level of the women. Methods: Elderly women (n = 376; age, 68.5 ± 14.1 years; height, 162.7 ± 5.5 cm; body mass, 71.2 ± 16.0 kg) who underwent ≥3 months of resistance training performed an initial week of familiarization and a second week of testing and retest, with a 48–72 h interval. Results: The results showed that Kappa indices ranged from 0.93 to 0.95, and the intraclass correlation coefficients were 0.99 for both the lower and upper limbs. In addition, minimal detectable changes were found that ranged between 1 and 3%, which means that changes lower than 1 kg could be detected. Conclusion: The present study confirms that the 1-RM test has high reliability and reproducibility in the elderly, for both upper and lower limbs.