Despite the fact that all new anticonvulsants have undergone extensive pharmacokinetic scrutiny prior to their introduction to the market, the opportunity to perform good prospective studies on their concentration-effect relationship has been largely missed, in some cases deliberately because therapeutic drug monitoring (TDM) is considered unfavourable for the marketing of a new drug. However, there are reasons to believe that TDM may play a useful role in maximising the therapeutic potential of new anticonvulsants. In fact, these drugs have a narrow therapeutic index, careful individualisation of dosage to optimise response is required, and inter- and intra-individual pharmacokinetic variability may translate into differences in dosage requirements. The wide interindividual variability in the serum concentrations at which therapeutic and toxic effects of these drugs are observed does not necessarily imply that TDM cannot be useful: indeed, a marked pharmacodynamic variability has also been reported for all the currently monitored older anticonvulsants.The new anticonvulsants which, based on their properties, are particularly attractive candidates for TDM include lamotrigine, topiramate, tiagabine, zonisamide and felbamate. However, in the absence on sound information on the target concentration ranges of these drugs, the routine concentration monitoring of these drugs cannot be recommended. Despite this, serial measurements of serum drug concentrations may be useful in selected patients, especially those suspected of poor compliance and those in whom pharmacokinetic changes caused by disease or administration of concomitant medication are anticipated. Even in the presence of marked interindividual pharmacodynamic variability, it is often possible to empirically determine the concentration at which each patient exhibits the best response, and apply that information in subsequent management. Prospective studies, using preferably a randomised concentration-controlled design, are necessary to better characterise concentration-effect relationships for these agents.