A blood sample containing CO in a glass vessel was heated in a thermostated water bath at various temperatures for varying lengths of time to compare changes in CO and water contents. With higher temperatures and longer exposure periods the degrees of thermocoagulation advanced and the contents of CO and water decreased. At the individual temperatures examined, changes of CO and water contents showed significant correlations. However, with different temperatures regressions differed significantly. The above was considered to show that the temperature, to which a blood sample had been exposed, must be known for successful application of the water content method to estimate pre-exposure CO content of heat-exposed blood. However, this requirement is considered difficult to be met.