The aim of the experiments was to investigate how manipulating the contrast of the signal and noise dots in a random dot kinematogram (RDK), influenced on motion coherence thresholds in adults with dyslexia. In the first of two experiments, coherent motion thresholds were measured when the contrasts of the signal and noise dots in an RDK were manipulated. A significantly greater processing benefit was found for the group with dyslexia than a control group when the signal dots were of higher contrast than the noise dots. However, a significant processing disadvantage was found for the group with dyslexia relative to the control group when the signal dots were of lower contrast than the noise dots. These findings were interpreted as supporting evidence for the noise exclusion hypothesis of dyslexia. In Experiment 2, the effect on coherent motion thresholds of presenting a cue that alerted observers to which stimuli, high or low contrast contained the signals dots was investigated. When the cue directed attention to low contrast signal dots presented in high contrast noise, coherent motion thresholds were only enhanced for the group with dyslexia. This manipulation produced equivalent coherent motion thresholds in the reader groups. In other conditions, the group with dyslexia had significantly higher coherent motion thresholds than the control group. It was concluded that adults with dyslexia who show evidence of a coherent motion deficit (37% of the dyslexia group in each experiment), have a specific difficulty in noise exclusion. This appears to occur as consequence of a sensory processing deficit in the magnocellular or dorsal stream. (Contains 2 tables and 4 figures.)