These authors contributed equally to this work. Abstract Established models of perceptual metacognition, the ability to evaluate our perceptual judgements, posit that perceptual confidence depends on the strength or quality of feedforward sensory evidence. However, alternative theoretical accounts suggest the entire perception-action cycle, and not only variation in sensory evidence, is monitored when evaluating confidence in one's percepts. Such models lead to the counterintuitive prediction that perceptual confidence should be directly modulated by features of motor output. To evaluate this proposal here we recorded electromyographic (EMG) activity of motor effectors while subjects performed a near-threshold perceptual discrimination task and reported their confidence in each response in a pre-registered experiment. A subset of trials exhibited subthreshold EMG activity in response effectors before a decision was made. Strikingly, trial-by-trial analysis showed that confidence, but not accuracy, was significantly higher on trials with subthreshold motor activation. These findings support a hypothesis that preparatory motor activity, or a related latent variable, impacts upon confidence over and above performance, consistent with models in which perceptual metacognition integrates information across the perception-action cycle.