We have compared estimation of alveolar carbon monoxide measured either as end-expired or mixed-expired carbon monoxide tension with two spectrophotometric estimations of venous blood carboxyhaemoglobin (COHb). Estimation of mixed-expired carbon monoxide proved more convenient for patients than the end-expired method, and this estimation of carbon monoxide exposure discriminated between non-smokers and smokers as well as measurements of COHb. The technique is bloodless, economical, and simple to perform. It should prove a valuable method for validating reported smoking habits of people attending smoking advisory centres and for epidemiological surveys of smoking.