Understanding drivers of biodiversity is a long-standing goal of basic and applied ecological research. In riverine systems, there remains a critical need to identify these drivers as efforts to manage and protect rivers grow increasingly desperate in the face of global change. We explored one commonly cited potential driver of riverine biodiversity, stream size (e.g., stream order, watershed area, width), using a systematic literature review paired with an analysis of broad-scale macroinvertebrate and fish communities. Of the 165 papers reviewed, we found mostly positive, but no universal, relationship between biodiversity and stream size despite inconsistent use of over 30 measures of stream size. One-third of studies failed to report explanatory mechanisms driving biodiversity-stream size relationships. Across over 4000 macroinvertebrate and fish samples from 1st-8th order streams in the contiguous USA, our analysis showed biodiversity (Shannon diversity, functional diversity, beta diversity) generally increased with measures of stream size. However, because of inconsistent and generally weak relationships between biodiversity and stream size across organismal groups, we emphasize the need to look beyond simple physical stream size measures to understand and predict riverine biodiversity, and strongly suggest that studies search for more mechanistic explanations of biodiversity patterns in lotic systems.