The residential sector is a major source of air pollutant emission inventory uncertainties. A nationwide field emission measurement campaign was conducted in rural China to evaluate the variabilities of realistic emission factors (EFs) from indoor solid fuel combustion. For a total of 1313 burning events, the overall average EFs (±standard deviation) of PM2.5 were 8.93 ± 6.95 and 7.33 ± 9.01 g/kg for biomass and coals, respectively, and 89.3 ± 51.2 and 114 ± 87 g/kg for CO. Higher EFs were found from burning of uncompressed straws, while lower EFs were found from processed biomass pellets, coal briquettes, and relatively clean anthracite coals. Modified combustion efficiency was found to be the most significant factor associated with variations in CO EFs, whereas for PM2.5, fuel and stove differences determined its variations. Weak correlations between PM2.5 and CO indicated high uncertainties in using CO as a surrogate for PM2.5. EFs accurately fit log-normal distributions, and obvious spatial heterogeneity was observed attributed to different fuel-stove combinations across the country. Emission estimation variabilities, which are determined by the interquartile ranges divided by the median values, were notably reduced when spatially resolved EFs were adopted in the inventory.