Both forest-floor litter and roots are known to be involved in the formation of soil organic carbon (SOC), but it remains elusive how the two contrasting types of plant carbon sources would differentiate in their effects on SOC. Here we used field-based soil mesocosms to study the responses of SOC components (i.e. mineral-associated organic C [MAOC], particulate organic C [POC], and dissolved organic C [DOC]) and pathways of production (i.e. plant-derived vs. microbial-derived C compounds) to multi-year treatments of forest-floor litter and roots in a temperate oak forest in central China. After eight years, the concentrations of both POC and MAOC were significantly higher in plots with presence of forest-floor litter than in those exclusive of both forest-floor litter and roots at the soil depth of 0–5 cm; roots predominantly affected POC but not the total SOC. The contributions from the forest-floor litter and roots to total SOC and its components were all non-additive; the plots with co-occurrence of forest-floor litter and roots had 49.7% less total SOC, 36.6% less POC, and 54.7% less MAOC, than the sum of the plots with only forest-floor litter or roots, at the soil depth of 0–5 cm. The concentrations of both lignin phenols and microbial biomass carbon (MBC) were significantly higher in plots with only forest-floor litter than in those exclusive of both forest-floor litter and roots down to a depth of 20 cm, and the plant-derived neutral sugars were significantly higher in plots with only forest-floor litter than in those exclusive of both forest-floor litter and roots at the soil depth of 0–10 cm. In contrast, the concentration of lignin phenols was significantly higher in plots with only roots than in those exclusive of both forest-floor litter and roots below the depth of 5 cm, and the concentrations of MBC and microbial-derived neutral sugars were higher in plots with only roots than in those exclusive of both forest-floor litter and roots at the soil depth of 0–10 cm. The plots with co-occurrence of forest-floor litter and roots had a significantly higher concentration of lignin phenols at all soil depths investigated, as well as higher concentrations of the plant- and microbial-derived neutral sugars and MBC at the soil depth of 0–5 cm, than those exclusive of both forest-floor litter and roots; the effects were all non-additive such that the plots with co-occurrence of forest-floor litter and roots had less plant- and microbial-derived C compounds than the sum of the values between the plots with only forest-floor litter and the plots with only roots. The contributions of amino sugars to SOC did not significantly respond to either the treatments with forest-floor litter and roots or soil depth, albeit variable MBC concentrations. In addition, our results show that the microbial-derived neutral sugars were positively correlated with plant-derived neutral sugars. Findings in this study provide empirical evidence on the primacy of forest-floor litter in SOC pool at the soil depth of 0–5 cm, and demonstrate the differential roles and contributions between forest-floor litter and roots in the pathways of SOC formation and spatial differentiation of carbon compounds.