To perform MRI cerebellum volumetry in patients exposed to phenytoin and to identify factors associated with cerebellar atrophy (CA). From 100 consecutive epilepsy patients we selected those with phenytoin use for more than 2 months and with MRI scan available for volumetric studies. We obtained cerebellar volumes corrected for total intracranial volume. Volumes below 2 standard deviations from the mean of control group were considered abnormal. We studied 56 patients (33 women). Mean age was 33.6 years and mean duration of epilepsy was 17.6 years. Mean daily dose of phenytoin was 301 mg. CA was detected in 20 (35.7%) patients. CA was not associated with frequent generalised seizures. CA correlated with duration of epilepsy (r=-0.34; P=0.01) and years of treatment with phenytoin (r=-0.48; P=0.001), but not with age and mean daily dosage of phenytoin (P>0.05). However, a multiple correlation analysis as well as a backward stepwise multiple regression analysis including all variables showed that only duration of treatment was significantly associated with CA (P=0.001). CA is frequently associated with long-term use of phenytoin. Although duration of epilepsy may have an influence in the CA, this is clearly less important than the time of exposure to phenytoin.12312-5