Human movement relies on noisy processes in neurons, muscle cells and sensory cells. Therefore, movements are variable and can never be exactly reproduced. The nervous system seems to exploit this movement noise for motor learning and specifically motor adaptation. However, a positive relation between movement noise and motor adaptation has not been consistently found in motor adaptation literature. Possibly, noise is comprised of distinct processes which contribute to motor adaptation in different ways. In Kalman filter theory, motor adaptation rate is calculated optimally from state noise and output noise, with state noise and adaptation rate positively correlated and output noise and adaptation rate negatively correlated. Therefore, if people learn (close) optimally from error, we would expect a similar relation. To investigate the relation between state noise, output noise and adaptation rate, we performed a visuomotor reaching adaptation experiment with a baseline and a perturbation block in 69 subjects. State noise, output noise and adaptation rate in the baseline and perturbation block were extracting using Bayesian fitting of a trial-to-trial state-space model. We found that adaptation rate in the perturbation block correlates positively with baseline state noise (r=0.27; 95%HDI=[0.05 0.50]) and negatively with baseline output noise (r= 0.41; 95%HDI=[ 0.63 0.16]). In addition, the steady-state Kalman gain calculated from baseline state and output noise correlated positively with adaptation rate in the perturbation block (r = 0.31; 95%HDI = [0.09 0.54]). Therefore, noise can be viewed both as a supporting factor for motor adaptation (state noise) and as a noise factor hampering optimal performance (output noise), and in order to understand the relationship of noise to learning, one must decompose noise into its constituent components.