Long-term surface cerebral blood flow (CBF) monitoring was performed to test the hypothesis that temporal lobe epileptogenicity is a function of epileptic cortical perfusion. Forty-three bitemporal 2-hour periictal CBF studies were performed in 13 patients. Homotopic regions of temporal cortex maintained interictal epileptic cortical hypoperfusion and nonepileptic normal cortical CBF. At 10 minutes preictus, a statistically significant, sustained increase in CBF was detected on the epileptic temporal lobe. Two minutes preictus, there was approximation of CBF in the epileptic and nonepileptic temporal lobes. Thereafter, electrocorticographic (ECoG) and clinical seizure onset occurred. The linear relationship between CBF in the two hemispheres (epileptic and nonepileptic) was the inverse of normal (y = -0.347x + 62.767, r = 0.470, df = 95, p < 0.05). The data indicated a direct linear correlation between epileptic cortical CBF and seizure interval (frequency-1), a clinical measure of epileptogenicity (r = 0.610, df = 49, p < 0.05). Epileptogenicity was also found to be a logarithmic function of the difference between nonepileptic and epileptic cortical perfusion (r = 0.564, df = 58, t = 5.20, p < 0.05). The results showed that progressive hypoperfusion of the epileptic focus correlated with a decreased seizure interval (increased epileptogenicity). Increased perfusion of the epileptic focus correlated with an increased seizure interval (decreased epileptogenicity). The fact that CBF alterations precede ECoG seizure activity suggests that vasomotor changes may produce electrical and clinical seizure onset.