Altered motor behaviour is commonly reported in Autism Spectrum Disorder, but the aetiology remains unclear. Here, we have taken a computational approach in order to break down motor control into different components and review the functioning of each process. Our findings suggest abnormalities in two areas--poor integration of information for efficient motor planning, and increased variability in basic sensory inputs and motor outputs. In contrast, motor learning processes are relatively intact and there is inconsistent evidence for deficits in predictive control. We suggest future work on motor abilities in autism should focus on sensorimotor noise and on higher level motor planning, as these seem to have a significant role in causing motor difficulties for autistic individuals.