We aimed to identify the physiological and muscle characteristics associated with mean power output in the final stage of a rowing incremental test in elite rowers. Twenty-two national level rowers (n = 9 female) completed the following assessments (i) a 7-stage incremental rowing ergometer test to determine V̇O2peak, maximal muscle deoxygenation and final incremental stage mean power output, (ii) strength and power, (iii) lean body mass (LBM) by dual-energy X-ray absorptiometry, (iv) lower extremity muscle volumes using MR imaging, (v) muscle architecture using diffusion tensor imaging; and (vi) estimated muscle typology via proton MR spectroscopy. Hierarchical multiple linear regression was used to identify the characteristics associated with final incremental stage mean power output with sex and LBM as covariates. Sex explained a substantial proportion of the variance in final incremental stage mean power output (R2 = 0.74; adjusted ΔR2 = 0.72; p textless 0.001), while inclusion of LBM further improved model fit (R2 = 0.87; ΔR2 = 0.13; adjusted ΔR2 = 0.13; p = 0.006). After accounting for sex and LBM, V̇O2peak demonstrated the largest independent association with final incremental stage mean power output (ΔR2 = 0.047; adjusted ΔR2 = 0.047; partial r = 0.60; p = 0.0048), followed by the estimated proportion of type II fibers (ΔR2 = 0.035; adjusted ΔR2 = 0.033; partial r = 0.52; p = 0.0188). Most other muscle characteristics explained minimal additional variance. Sex and muscle characteristic interaction terms indicated no detectable sex-specific differences in this small cohort. These findings highlight the diversity in the underpinning muscle characteristics and may inform personalized interventions targeted at improving rowing performance.