Smoking can be detected by the determination of cotinine in urine. We compared the performance of an automated modification of the 'König' reaction adapted to a centrifugal analyzer with an automated commercial fluorescence polarization immunoassay (TDX system). In the latter assay, cotinine, as the primary metabolite of nicotine, can be measured with high specificity. In contrast, the 'König' reaction also detects nicotine metabolites other than cotinine by a group colour reaction. Analysis speed of the 'König' reaction was about 66 samples/h with a detection limit 2 S.D. above the mean value of urine samples of non-smokers. Analysis speed of the TDX system was 41 samples/h. The coefficient of variation (C.V.) of both methods in smokers' urine was 8.6% ('König' reaction) vs. 3.4% (TDX system) in the high range and 16.4% vs. 9.5% in the low range. In a controlled, prospective study recruiting 86 cigarette-smoking volunteers, 83.7% were correctly classified as being smokers by both systems, 13.9% were classified as smokers by the 'König' reaction only and 2.4% were misclassified as non-smokers by both systems. Thus, the sensitivity of the 'König' reaction seems to be higher than in the TDX system (97.6% vs. 83.7%). Of 33 non-smoking individuals, 81.8% were correctly classified as non-smokers by both systems, 18.2 were misclassified as smokers by the 'König' reaction and no person was misclassified by the fluorescence polarization immunoassay. Thus, the specificity of the TDX system seems to be higher than that of the 'König' reaction (100% vs. 81.8%). We conclude that both systems are applicable to detect individuals who smoke regularly by simple urine testing. The higher specificity of the TDX system is outweighed by the higher sensitivity of the 'König' reaction at much lower cost.