The correlation between cotinine concentrations in a single specimen of urine and in serum collected on the same occasion from 279 male smokers was 0.83. This was significantly increased, to 0.91, by adjusting the urinary cotinine levels for urinary creatinine concentration to take account of variations in urinary dilution between people. The adjustment used was based on the observed regression relationship between urinary cotinine and urinary creatinine concentrations. Expressing urinary cotinine values as a ratio to urinary creatinine, which has been used as a method of adjustment by others, did not improve the correlation between serum and urinary cotinine levels. The method of adjusting urinary cotinine for urinary creatinine is, therefore, important. The principle of such adjustment should apply not only to cotinine but also to other urinary biochemical measurements.