This study was concerned with estimating endogenous temperature rhythms without imposing sleep deprivation. The aim of Experiment 1 was to quantify the masking effect on the circadian temperature rhythm in a group of 18 healthy young subjects (8 women and 10 men, ages 19-29 years). Temperature data collected under a 36-h wakeful bed rest protocol were used as a marker of the endogenous component of the rhythm ("unmasked rhythm"), and temperature data collected under 24 h of a normal nycthemeral routine (immediately before the bed rest protocol) were used as the "masked" rhythm. An algorithm to "demask" the temperature data collected under the nycthemeral condition was then developed, based on the differences observed between the temperature data collected under wakeful bed rest and nycthemeral conditions. The consistency of the demasking technique was tested in Experiment 2, using the same parameters on a group of 19 healthy elderly subjects (8 women and 11 men, ages 78-88 years) who also had experienced both nycthemeral and wakeful bed rest conditions. The demasking technique was evaluated both by comparing nycthemeral, demasked, and unmasked temperature rhythms themselves and by comparing individual estimates of circadian phase and amplitude that had been gleaned from them. In comparison to the unmasked condition, the nycthemeral condition showed lower mean nighttime temperature, earlier mean phase estimates, and higher mean amplitude estimates in both young and elderly subjects. Following application of the demasking procedure to the nycthemeral temperature data, mean demasked temperature curves were closely comparable to mean unmasked temperature curves in both young and elderly subjects. Phase and amplitude estimates derived from the demasked temperature data also were highly comparable to those in the unmasked conditions. Thus, this demasking procedure appears to be a useful tool in estimating the endogenous temperature rhythm and appears to work equally well for young and elderly subjects.