Fire experiments were conducted in small scale compartment models under a forced or natural ventilation. Four liquid fuels of methanol, ethanol, n-hexane and benzene, which are considered to represent thermal decomposition products of polymers, were burned in the fire. Concentrations of CO, O2 and CO2 just below the ceiling of the model chamber were determined, and the temperature of the fire gas also near the ceiling was measured. For the comparison purpose, these liquid fuels were also burned in a horizontal tube furnace to determine those gas concentrations. Results are summarized as follows. I. For the chamber experiments, the maximum CO concentration was found to be about 10% for methanol and ethanol, and about 5 % for n-hexane and benzene at less than 1/4 of stoichiometric air supply rate. For the furnace experiments, the maximum CO concentration was 2 to 4 times higher than those for the chamber experiments, and differed rather little according to the fuel. Low temperature is probably one of the reasons for the low maximum CO concentrations for the chamber experiments. Therefore, the maximum CO concentration will possibly rise closer to those for the furnace experiments if the temperature is raised by scaling up the model. 2. The air supply rate for the maximum CO concentration did not always agree with that for the maximum CO generation. Mostly the latter was larger than the former. 3. Generally CO concentration was high at less than 2 % of oxygen concentration. 4. The results of the furnace experiments indicate that the maximum CO generation generally increases with increasing carbon content of the fuel. 5. When the possibility of death due to CO poisoning, oxygen deficiency or heat was considered on the basis of the data of the chamber experiments under natural ventilation and the lethal levels of CO, O2 and heat for 5 minute exposure, it was found that CO poisoning and heat are far more likely to cause death than oxygen deficiency.