The one-repetition maximum (1RM) test is considered the gold standard for assessing muscle strength in non-laboratory situations. Since most previous 1RM reliability studies have been conducted with experienced young participants, it is unclear if acceptable test-retest reliability exists for untrained middle-aged individuals. This study examined the reliability of the 1RM strength test of untrained middle-aged individuals. Fifty-three untrained males (n=25) and females (n=28) aged 51.2+/-0.9 years participated in the study. Participants undertook the first 1RM test (T1) 4-8 days after a familiarisation session with the same exercises. 1RM was assessed for seven different exercises. Four to eight days after T1, participants underwent another identical 1RM test (T2). Ten weeks later, 27 participants underwent a third test (T3). Intraclass correlation coefficients (ICC), typical error as a coefficient of variation (TEcv), retest correlation, repeated measures ANOVA, Bland-Altman plots, and estimation of 95% confidence limits were used to assess reliability. A high ICC (ICC>0.99) and high correlation (r>0.9) were found for all exercises. TEcv ranged from 2.2 to 10.1%. No significant change was found for six of the seven exercises between T1 and T2. Leg press was slightly higher at T2 compared to T1 (1.6+/-0.6%, p=0.02). No significant change was found between T2 and T3 for any exercise. 1RM is a reliable method of evaluating the maximal strength in untrained middle-aged individuals. It appears that 1RM-testing protocols that include one familiarisation session and one testing session are sufficient for assessing maximal strength in this population.