The relationship between a dose of phenytoin and the resultant serum concentration is difficult to predict, and numerous dosing methods have been developed to quantify the dose required to achieve a specific concentration. This review brings up to date the earlier article in the Journal regarding predictive algorithms, various pharmacokinetics-based dosing techniques and Bayesian feedback methods for phenytoin dosing. The latest data support the original conclusions that dosing methods for phenytoin which incorporate an individualised approach or Bayesian principles tend to offer results superior to those from predictive algorithms. Bayesian methods have the additional advantage of using only 1 serum concentration, obtained under either steady-state or non-steady-state conditions. There is still a need for future investigations that include prospective evaluations of predictive performance and cost-effectiveness data.