Abstract. The quantification and mapping of surficial seabed sediment organic carbon has wide-scale relevance across marine ecology, geology and environmental resource management, with carbon densities and accumulation rates being a major indicator of geological history, ecological function, and ecosystem service provisioning, including the potential to contribute to nature-based climate change mitigation. While global mapping products can appear to provide a definitive understanding of the spatial distribution of sediment carbon, there is inherently high uncertainty when making estimates at this scale. Finer resolution national maps which utilise targeted data syntheses and refined spatial data products are therefore vital to improve these estimates. Here, we report a national systematic review of data on organic carbon content in seabed sediments across Canada and combine this with a synthesis and unification of best available data on sediment composition, seafloor morphology, hydrology, chemistry, geographic setting and sediment mass accumulation rates within a machine learning mapping framework. Predictive quantitative maps of mud content, sediment dry bulk density, and organic carbon content, density and accumulation, were each produced along with cell specific estimates of their 95 % confidence interval (CI) bounds at 200 m resolution across 4,489,235 km2 of the Canadian continental margin (92.6 % of the seafloor area above 2,500 m). Fine-scale variation in carbon stocks was identified across the Canadian continental margin, particularly in the Pacific and Atlantic Ocean regions. Carbon accumulation was predicted to be concentrated in coastal areas, with the highest rates in the Gulf of St Lawrence and Bay of Fundy. Overall, we estimate the standing stock of organic carbon in the top 30 cm of surficial seabed sediments across the Canadian shelf and slope to be 10.7 Gt (95 % CI 6.6 – 16.0 Gt), and accumulation at 4.9 Mt per year (95 % CI 2.6 – 9.3 Mt y-1). Increased in-situ sediment data collection and higher precision in spatial environmental data-layers could significantly reduce uncertainty and increase accuracy in these products over time.