Para-cresol (4-methylphenol) is a volatile phenolic compound which is retained in chronic renal failure. Several recent studies suggest that p-cresol interferes with various biochemical and physiological functions at concentrations currently observed in uremia. Only a few methods are available for the determination of p-cresol concentration in serum. In addition, these methods have only been used for the determination of total p-cresol. In particular, the evolution of free (non-protein bound) p-cresol is of concern, because conceivably this is the biologically active fraction. The concentration of free p-cresol, is, however, markedly lower than that of total p-cresol, in view of its important protein binding. We report a method enabling the measurement of total and free p-cresol concentration in serum of healthy controls and uremic patients. Deproteinization, extraction and HPLC procedure are efficient, without interference of other protein bound ligands and/or precursors of p-cresol or phenol. By means of spiking experiments, the measurement of the UV absorbance over the 200-400 nm wavelength range, and capillary gas chromatography-mass spectrometry, the considered compound is identified as p-cresol. With a fluorescence detection at 284/310 nm as extinction/emission wavelengths the detection limit of p-cresol is 1.3 micromol/l (0.14 microg/ml). Recovery of added p-cresol to normal serum is 95.4+/-4.1%. For free p-cresol and total p-cresol determinations, intra-assay and day-to-day variation co-efficients are 3.2%, 4.2%, 6.9% and 7.3%, respectively. Compared to healthy controls, the serum p-cresol levels are 7-10 times higher in continuous ambulatory peritoneal dialysis patients (CAPD), uremic outpatients, and hemodialysis patients: 8.6+/-3.0 vs. 62.0+/-19.5, 87.8+/-31.7 and 88.7+/-49.3 micromol/l (0.93+/-0.32 vs. 6.70+/-2.11, 9.49+/-3.43, and 9.60+/-5.30 microg/ml) (p<0.05), respectively. The difference is even more important if free p-cresol is considered. This corresponds to a decreased protein binding in uremic patients. We conclude that the present method allows an accurate measurement of both total and free p-cresol, and that the measured concentrations in uremia are in the range which may cause biochemical alterations.