The nature and occurrence of remission, and conversely, pharmacoresistance following epilepsy treatment is still not fully understood in human or veterinary medicine. As such, predicting which patients will have good or poor treatment outcomes is imprecise, impeding patient management. In the present study, we use a naturally occurring animal model of pharmacoresistant epilepsy to investigate clinical risk factors associated with treatment outcome. Dogs with idiopathic epilepsy, for which no underlying cause was identified, were treated at a canine epilepsy clinic and monitored following discharge from a small animal referral hospital. Clinical data was gained via standardised owner questionnaires and longitudinal follow up data was gained via telephone interview with the dogs' owners. At follow up, 14% of treated dogs were in seizure-free remission. Dogs that did not achieve remission were more likely to be male, and to have previously experienced cluster seizures. Seizure frequency or the total number of seizures prior to treatment were not significant predictors of pharmacoresistance, demonstrating that seizure density, that is, the temporal pattern of seizure activity, is a more influential predictor of pharmacoresistance. These results are in line with clinical studies of human epilepsy, and experimental rodent models of epilepsy, that patients experiencing episodes of high seizure density (cluster seizures), not just a high seizure frequency pre-treatment, are at an increased risk of drug-refractoriness. These data provide further evidence that the dog could be a useful naturally occurring epilepsy model in the study of pharmacoresistant epilepsy.