Oral temperature has long been used as a basis for clinical decisions. The ingestion of cold liquids such as iced water, which are frequently found at patients' bedsides or in clinic waiting areas, has been believed to potentially lower oral temperature readings. The degree and duration of this effect, however, has not been well established. This study examined the effect of iced water on oral temperature as measured with a standard electronic thermometer. Ninety-two afebrile adults were stratified into three age groups. Their temperatures were measured every two minutes for ten minutes to establish a baseline. Subjects then drank 9 oz of iced water over two minutes and their temperatures were again taken every two minutes for the next half hour. The overall mean initial change was -2.4F (P less than .001; range, +0.1 to -8.2 degrees). Although no differences existed at baseline, temperature decreases and recovery time were strongly related to age. Mean decreases were -3.1F for persons 60 years or older, -2.4F for those between 40 and 59 years, and -1.6F for those less than 40 years of age (P = .0001). Median times for return to baseline temperature were 26, 18, and 14 minutes, respectively. On the basis of these findings it is recommended that oral temperature not be measured immediately after iced water ingestion. It is further recommended that oral temperature readings be delayed at least 15 minutes in people under the age of 40 years, 20 minutes in those who are 40 to 59 years, and 30 minutes in persons aged 60 and older.