In the course of experiments on focal epilepsy in rats, we have recognized that there are no adequate models of subacute focal epilepsy in rodents. We have, therefore, reevaluated a previously described rat model that reliably generates subacute seizures over 2-3 weeks. After implantation of a short length of cobalt wire into the left motor cortex, the animals are monitored by standard EEG over the next 3 weeks. They develop three seizure types: 1. Simple partial seizures with contralateral clonic jerks, lasting 17.9 +/- 46.4 min; these seizures were characterized by repetitive single spikes; 2. Secondarily generalized seizures, lasting 34.5 +/- 19.0 s; and 3. Complex partial seizures with a paroxysmal EEG, lasting 39.6 +/- 55.5 s. Post mortem brains were imaged using standard magnetic resonance techniques, after removal of the ferromagnetic cobalt wire. There was a localized loss of the MR signal that differed by pulse sequence, indicating spread of the ferromagnetic cobalt into the brain tissue. The image disruption caused by the cobalt was quite abrupt, indicating a sharp cobalt concentration gradient. However, we saw no evidence of widespread cerebral injury. The unilateral cobalt wire model generates less frequent, but more persistent seizures than seen in most acute, focal models. The ferromagnetic signal present, even after wire removal, indicates that metallic cobalt leaches into the cortex and may be responsible for generating the seizures. This model should be useful for testing new therapies for neocortical epilepsy.