Small headwater streams can mobilize large amounts of terrestrially derived dissolved organic matter (DOM). While the molecular composition of DOM has important controls on biogeochemical cycles and carbon cycling, how stationary landscape metrics affect DOM composition is poorly understood, particularly in relation to non-stationary effects from hydrological changes across seasons. Here, we apply a combination of Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR-MS) and absorbance spectroscopy to characterize stream DOM from 13 diverse watersheds across the central Canadian boreal forests and statistically relate DOM compositional characteristics to landscape topography and hydrological metrics. We found that watershed runoff across different surface physiographies produced DOM with distinctly different chemical compositions related to runoff pH. Specifically, streams in sandy soil watersheds contained more abundant aromatic, nitrogenated and sulfurized fractions of DOM, likely due to a combination of lower soil capacity to absorb DOM than other soil types and high conifer forest coverage that generated acidic litterfall in more sandy watersheds. In contrast, streams with more neutral pH in watersheds with shallow soils had DOM resembling low oxidized phenolic molecules mainly due to increased brush/alder and deciduous vegetation coverage in relatively steeper watersheds. However, as precipitation and flows increased in the fall, the overall water chemistry of streams became more similar as runoff pH increased, the overall chemical diversity of DOM in streams decreased, and stream DOM resembled fresher, lower molecular weight lignin material likely originating from freshly produced leaf litter. Together, our findings show that during hydrologically disconnected periods, pH and landscape characteristics have important controls on the mobilization of aromatic DOM but that many landscape-specific characteristics in the Canadian boreal forest are less influential on DOM processing during wetter conditions where chemically similar, plant-derived DOM signatures are preferentially mobilized. These findings collectively help predict the composition of DOM across diverse watersheds in the Canadian boreal to inform microbial and contaminant biogeochemical processes in downstream ecosystems.