Experimental modeling in the C-O-H, C-O-H-N, and peridotite-C-O-H-N systems, combined with analyses of fluid inclusions in natural diamonds, is used to reconstruct the compositions of fluids that can be stable in the reduced mantle. Hydrocarbons (HCs) in the upper mantle can form either by reactions of carbonates with iron/wüstite and water or by direct hydrogenation of carbon phases (graphite, diamond, and amorphous carbon) interacting with reduced fluids. Carbon required for the formation of HCs can come from diamond, graphite, or carbonates. Mainly light alkanes are stable at the mantle pressures and temperatures in the C-O-H and C-O-H-N systems as well as in the peridotite-fluid system under ultrareduced to moderately reduced redox conditions at the oxygen fugacity from -2 to +2.5 lg units relative to the IW (Fe-FeO) buffer. Some oxygenated HCs can be stable in fluids equilibrated with carbonate-bearing peridotite. Ammonia and, to a lesser degree, methanimine (CH3N) are predominant nitrogen species in reduced fluids in the conditions of the subcratonic lithosphere or the Fe0-bearing mantle. The presence of HCs as common constituents of reduced mantle fluids is supported by data on inclusions from natural diamonds hosted by kimberlites of the Yakutian province and from placer diamonds of the northeastern Siberian craton and the Urals. Fluid inclusions have minor amounts of H2O, methane, and other light alkanes but relatively high concentrations of oxygenated hydrocarbons, while the H/(H + O) ratio varies from 0.74 to 0.93. Hydrocarbon-bearing fluids in some eclogitic diamonds have high CO2 concentrations. Also, the fluid inclusions have significant percentages of N2 and N-containing species, Cl-containing HCs, and S-containing compounds. Both the experimental results and the analyses of fluid inclusions in natural diamonds indicate that HCs are stable in the upper mantle conditions. The set of hydrocarbons, mainly light alkanes, might have formed in the mantle from inorganic substances. Further research should focus on the causes of the difference between experimental and natural fluids in the contents of methane, light alkanes, oxygenated hydrocarbons, and water and on the stability of N-, S-, and Cl-containing fluid components.