OBJECTIVES: Phenytoin is an anti-epileptic drug that has a narrow therapeutic index, and therefore requires therapeutic drug monitoring. Only the free fraction is pharmacologically active, and in some cases, accurate determination of the free phenytoin concentration may be essential to prevent phenytoin toxicity. Although it is possible to measure free phenytoin concentration, often only the total concentration is measured, with equations used to estimate the free fraction. Several equations are quoted in the literature with no overall consensus with regard to accuracy. This study aimed to assess the correlation between total and free phenytoin in a mixed patient population, and to compare the accuracy of several different equations used to estimate the free phenytoin concentration. METHODS: Fifty-one serum samples were analysed for total phenytoin, free phenytoin and albumin. The measured free phenytoin concentrations were compared against those estimated using five selected equations, identified through a literature search. RESULTS: This study showed poor correlation between the total and measured free phenytoin concentrations, and between the estimated and measured free concentrations. The overall correlation was concentration-dependent, but a correction factor could not be applied to improve the accuracy consistently. The equations assessed showed wide variability between the estimated and measured free phenytoin concentrations, with several showing a clinically significant negative bias when compared with the measured free fraction. DISCUSSION: This study highlights the disparity of the free phenytoin concentrations generated by the equations. Underestimation of free phenytoin concentrations using these equations may result in phenytoin toxicity, bringing into question the safety of using calculated values for patient management in place of physical measurement of free phenytoin concentration by ultra-performance liquid chromatography tandem mass spectrometry.