Parafoveal-on-foveal priming refers to the presentation of an item (the prime) in parafoveal vision followed by the presentation of an item (the target) in foveal vision. In natural reading, the 'parafoveal preview benefit' subserves fluent reading as, e.g., reading times increase when such information is not available. Yet, the neural correlates of reading are mostly studied with foveally presented stimuli and little is known of this parafoveal influence. Here, we used complementary information from a behavioral study and a magnetoencephalography experiment to clarify the relationship between parafoveal-on-foveal and foveal priming. Unlike foveal priming, parafoveal-on-foveal priming was present only at short prime-to-target delay (<100 ms). Behaviorally, the parafoveal priming effect was influenced by the prime visual field (left/right) and target lexical type (word/non-word), suggesting emphasis on perceptual analysis for LVF primes and on conceptual analysis for RVF primes. At the neural level, the overall sequence of activation was similar for foveal and parafoveal primes followed by foveal word targets, but the priming effects were bilateral for foveal primes versus left-lateralized for RVF primes. No neural effects of priming appeared for LVF primes, in line with the RVF preference imposed by the Western writing system. These results highlight the role of the left hemisphere in linguistic analysis and point out possible limitations of foveal stimulus presentation for drawing conclusions about natural reading.