Emission of smoke from various fire sizes and fuels has been examined using data from the literature. The data used for the examination were for the fully ventilated combustion of the mixed fuels of all compositions and for non-mixed single hydrogen atom containing fuels, without highly fire-retarded compositions. The combustion data used were for the smaller (0.008 m2 in area) and larger (0.911-m2 area) pool-like configuration. The data were also used for the combustion of fuels in vertical wall-like configuration consisting of single panel (0.31-m long and 0.10-m wide) and parallel panels (each panel 2.4- m long and 0.60-m wide separated by 0.30-m and 4.9-m long and 1.1-m wide separated by 0.53-m). The examination of these data show that the average smoke emission rate, S G ?? , depends on the fire size, expressed as the average chemical heat release rate, ch Q ?? , and the combustion chemistry expressed as the ratio of the yields or emission rates of CO to CO2 (yCO/CO2 or G??CO / G??CO2 respectively). Correlation is found between S G ?? and Qch ( yCO / yCO2 ) ?? or Qch (GCO / GCO2 ) ?? ?? ?? to the power 0.96 with a correlation constant of 0.053. The correlation limits are in the range of 0.0008 to 8 g/s for S G ?? and 0.015 to 100 kW for Qch (GCO / GCO2 ) ?? ?? ?? . The correlation holds for particulate dominated smoke and for fuels with non-particulate dominated smoke in the presence of H and OH atoms provided by other fuels or by the ignition source, such as a hydrocarbon gas burner. Deviation of the experimental data from the correlation appears to be due to the unaccountability of the non-particulates by the optical technique used to measure the particulates. A new technique that could measure both particulates and non-particulates in smoke would make the correlation to hold for all types of fuels and for all types of fire conditions including ventilation. Such a generalized correlation would be useful for smoke modeling and for the smoke hazard assessment.