This thesis reports an investigation of the component processes underlying reading fluency. A current controversy in reading research is whether reading ability and development is determined solely by linguistic processes such as phonological (sound-based) skill, or whether it can also be influenced by non-linguistic processes such as visual processing of print, attention and general timing capacity. One way of addressing this problem is to investigate naming speed; the speed with which objects, colours, letters or digits can be named. Letter naming in particular represents a ‘microcosm’ of the processes required for fluent reading. The experiments in this thesis systematically manipulate the letter naming process to investigate a) what determines naming-speed performance and b) which processes, when aberrant, cause slower naming in dyslexic readers. Results suggest that non-dyslexic readers are better able to process multiple letter items simultaneously than dyslexic readers. Further, we find evidence suggesting a strong role for phonological retrieval in determining naming latencies and contributing to the naming-speed deficit. We also identify a strong attentional component and a role for visual processing in naming speed which discriminates dyslexic and non-dyslexic reading groups. The results support models emphasising the multi-componential nature of reading fluency and suggest that naming deficits in dyslexia reflect processing difficulties in non-phonological, as well as phonological domains.