The present thesis examines the processing of numerical magnitude using a series of Stroop variants existing paradigms are modified and new ones devised. The Stroop effect, observed when numerical magnitude is the task-irrelevant dimension, has been widely used as an indicator for autonomous processing. However, it is argued that this effect does not provide a sensitive enough measure for the degree of information processing. Instead, current findings have demonstrated that the reversed numerical distance effect observed during physical size comparison of digits (when numerical magnitude is the task- irrelevant dimension) can be used as a reliable measure for refined autonomous processing. Neuroimaging data are consistent with this proposal. Others factors which influence numerical magnitude processing, such as effects of writing system and familiarity, are also examined. Findings are discussed with respect to existing theoretical accounts. In addition to the relative strength and speed of processing, stimulus discriminability and familiarity also contribute to a more comprehensive understanding of number comparison. The autonomous property of numerical magnitude processing during parity judgement and numerosity matching is also investigated. Current findings indicate that numerical magnitude processing is, in most cases, refined. Experimental designs have significant contributions as to whether refined processing can be elicited. Moreover, the present thesis provides evidence supporting the ideas that (1) representations of numerical magnitude follow the principle of cardinality and are therefore linear, and (2) numbers evoke discrete representations which are distinct from continuous presentations evoked by other quantitative dimensions such as physical size. Future research directions are suggested, with an emphasis on non-symbolic stimuli which can potentially be used as diagnostic tools for dyscalculia with young children, on the assumption that the aetiology of this developmental condition rests on a more general quantity processing deficit.