Of the 9 new anticonvulsants that have been marketed recently in the UK or US, a number appear to have either adverse or beneficial effects on behaviour. There is now a considerable database of information, in terms of the number of patients treated and/or the number of published reports, on vigabatrin, lamotrigine, gabapentin and topiramate. Oxcarbazepine has been available in some centres for several years and there is extensive experience with the drug in Scandinavia. It appears that the profile of adverse and beneficial effects is similar to that of carbamazepine. Behavioural effects have probably been greatest with vigabatrin, with psychosis, depression and other behavioural problems recorded, but the use of this drug has been limited because of the concern about visual field constriction. The cognitive and behavioural effects of topiramate have caused concern, but these may be much less of a problem if lower starting dosages and escalation rates are used. Psychosis and depression have been associated with topiramate, as they have with another carbonic anhydrase inhibiting drug, zonisamide. Although zonisamide has been used for many years in Japan and Korea, experience elsewhere with this drug is currently very limited. Gabapentin seems to be less associated with adverse behavioural effects than some of the other new anticonvulsant drugs. The reports of behavioural disturbance with gabapentin in children may be related to dose escalation. Behavioural disturbance as a direct result of lamotrigine seems to be uncommon, although indirect effects on behaviour, through the so-called 'release phenomenon' from improved seizure control and consequent ability to misbehave, can occur. Positive behavioural effects have been described with several of the new anticonvulsants, particularly gabapentin, lamotrigine and oxcarbazepine; all of these drugs may have mood-levelling effects that could be of value in treating affective disorders. The information on tiagabine and levetiracetam is too limited to allow any firm conclusions to be drawn with regard to positive or negative behavioural effects. When interpreting reports of behavioural changes with anticonvulsants, it is important to avoid attributing the effect to the drug when one or more of the other multiple causes of behavioural disturbance in people with epilepsy may be responsible or when an indirect effect such as 'forced normalisation' may be the cause. Many of the published studies are retrospective and unblinded rather than double-blind, placebo-controlled, prospective trials, implying that much of the data must be interpreted with caution at this stage.