The human definitions of epilepsy and seizure classification were applied rigidly to epileptic dogs to investigate whether the distribution of the seizure types and epilepsies of dogs is comparable to that of human beings. Sixty-three dogs were referred because of recurrent (>2) epileptic seizures. Only dogs without previous or ongoing antiepileptic treatment were included. All dogs had a physical and neurologic examination and blood work that included a CBC and a biochemical profile. All owners were asked to complete a questionnaire, focusing on seizure development. In addition, video recordings of suspected seizure episodes were analyzed if available. In the majority of dogs where an intracranial lesion was suspected, a computerized tomography scan was performed. Sixty-five percent of the dogs experienced partial seizures with or without secondary generalization and 32% exhibited primary generalized seizures; in 3% of the dogs the seizures could not be classified. Twenty-five percent of these cases were classified as idiopathic, 16% as symptomatic, and 45% as cryptogenic epilepsy; in 14% of these a classification was not possible. Applying human definitions, the distribution of seizure types and epilepsy classifications in these dogs differed widely from those in previous reports of canine epilepsy, where generalized seizures and idiopathic epilepsy were most frequently reported. However, our findings are consistent with the results of several large studies of human epilepsy patients. In dogs with epilepsy, closer attention must be given to the detection of a partial onset of seizures. In this study, detailed questioning of the owners and when possible analysis of video recorded seizures, proved to be sufficient for diagnosing seizures with a partial onset in a significant number of dogs. Partial onset of seizures may be an indication of underlying cerebral pathology. Some adjustments of veterinary epilepsy terminology are suggested.