Silicon (Si) is not classified as an essential plant nutrient, and yet numerous reports have shown its beneficial effects in a variety of species and environmental circumstances. This has created much confusion in the scientific community with respect to its biological roles. Here, we link molecular and phenotypic data to better classify Si transport, and critically summarize the current state of understanding of the roles of Si in higher plants. We argue that much of the empirical evidence, in particular that derived from recent functional genomics, is at odds with many of the mechanistic assertions surrounding Si’s role. In essence, these data do not support reports that Si affects a wide range of molecular-genetic, biochemical and physiological processes. A major reinterpretation of Si’s role is therefore needed, which is critical to guide future studies and inform agricultural practice. We propose a working model, which we term the ‘apoplastic obstruction hypothesis’, which attempts to unify the various observations on Si’s beneficial influences on plant growth and yield. This model argues for a fundamental role of Si as an extracellular prophylactic agent against biotic and abiotic stresses (as opposed to an active cellular agent), with important cascading effects on plant form and function.