Castillo, DC, Strand, KL, Oh, J, Eltoukhy, M, Totillo, MC, and Signorile, JF. The development of a regression model to predict object transfer power in older adults. J Strength Cond Res 34(11): 3086-3093, 2020-Declines in mechanical power affect independence in older adults; however, no practical clinical method exists to assess peak power (PP) and average power (AP) during transfer tasks in this population. Therefore, predictive models were developed and validated during performance of an existing transfer assessment, the gallon-jug shelf-transfer (GJST) test. Twenty-one independent-living older women (age = 74.8 ± 5.6) and men (age = 73.8 ± 4.5) participated in the development phase. Peak power and AP were computed using data from three-dimensional movement analyses while subjects moved five 1-gallon jugs from a low (0.59 m) to high shelf (1.43 m) as quickly as possible. Stepwise linear regression models using test duration, subject height, and age produced equations with high R values for PP (0.655-0.701; p < 0.0001) and AP (0.703-0.759; p < 0.0001). However, regression and Bland-Altman analyses showed models with declining agreement as power values increased. Therefore, analyses were repeated using a quadratic model, showing stronger predictive capacity (PP: R = 0.786; AP: R = 0.878; p < 0.0001). Validity of this model was evaluated in a second sample of 30 older women (age = 69.9 ± 4.7) and men (age = 70.7 ± 4.3). Computed GJST AP and PP were compared with power produced during dominant arm chest press and distance during the seated medicine ball throw. Generalizability of these models was also confirmed through cross-validation analyses. Our results indicate mechanical power in older adults can be accurately measured using the quadratic model, with test duration as the independent variable. This model provides a simple, safe, and inexpensive assessment tool that can be administered by clinicians to improve diagnostic and therapeutic processes in older persons.