The chronic treatment with phenytoin or the acute intoxication by this drug may cause permanent cerebellar injury with atrophy of cerebellum vermis and hemispheres, which can be detected by neuroimaging studies. The aim of the present study was to investigate the correlation between the dosage and duration of treatment with phenytoin and the occurrence of cerebellar atrophy. Sixty-six patients were studied and had their tomographies analyzed for cerebellar atrophy. Of the 66 patients studied, 18 had moderate/severe atrophy, 15 had mild atrophy and 33 were considered to be normal. The patients with moderate/severe atrophy were those with higher exposure to phenytoin (longer duration of treatment and higher total dosage) showing statistically significant difference when compared to patients with mild atrophy or without atrophy (p=0. 02). Further, the patients with moderate/severe atrophy had serum levels of phenytoin statistically higher than those of patients with mild atrophy or without atrophy (p = 0.008). There was no association between other antiepileptic drugs dosage or duration of treatment and degree of cerebellar atrophy. We also found that older patients had cerebellar atrophy more frequently, indicating that age or duration of the seizure disorder may also be important in the determination of cerebellar degeneration in these patients. We conclude that although there is a possibility that repeated seizures contribute to cerebellar damage, long term exposure to phenytoin, particularly in high doses and toxic serum levels, cause cerebellar atrophy.