PurposeJoint angle is a significant determinant of neuromuscular and metabolic function. We tested the hypothesis that previously reported correlations between knee-extensor torque complexity and metabolic rate ( ṁVO2 ) would be conserved at reduced joint angles (i.e. shorter muscle lengths).MethodsEleven participants performed intermittent isometric knee-extensor contractions at 50% maximum voluntary torque for 30 min or until task failure (whichever occurred sooner) at joint angles of 30º, 60º and 90º of flexion (0º = extension). Torque and surface EMG were sampled continuously. Complexity and fractal scaling of torque were quantified using approxi-mate entropy (ApEn) and detrended fluctuation analysis (DFA) α. ṁVO2 was determined using near-infrared spectroscopy.Results Time to task failure/end increased as joint angle decreased (P < 0.001). Over time, complexity decreased at 90º and 60º (decreased ApEn, increased DFA α, both P < 0.001), but not 30º. ṁVO2 increased at all joint angles (P < 0.001), though the magnitude of this increase was lower at 30º compared to 60º and 90º (both P < 0.01). There were significant correlations between torque complexity and ṁVO2 at 90º (ApEn, r = − 0.60, P = 0.049) and 60º (ApEn, r = − 0.64, P = 0.035; DFA α, ρ = 0.68, P = 0.015).ConclusionThe lack of correlation between ṁVO2 and complexity at 30º was likely due to low relative task demands, given the similar kinetics of ṁVO2 and torque complexity. An inverse correlation between ṁVO2 and knee-extensor torque com-plexity occurs during high-intensity contractions at intermediate, but not short, muscle lengths.