The potent NMDA receptor antagonist, Conantokin-G (CGX-1007), a snail peptide, has an 8-h therapeutic window in rat focal cerebral ischemia. We hypothesized that the mechanism of neuroprotection is the inhibition of 'secondary phase' peri-infarct depolarizations (PIDs), recently shown to recur 6--24 h post-reperfusion. Rats were implanted with intrathecal (i.t.) catheters for drug delivery and DC-compatible electrodes for continuous PID monitoring and subjected to transient (2 h) middle cerebral artery occlusion. Four groups were studied. In two groups (C(40)C and C(20)C), continuous infusion of CGX--1007 was administered over 8--24 h post-occlusion at 0.1 microg/h (0.04 nmol/h) following either a 40- or 20-nmol bolus dose at 8 h. Another group (C(40)S) received the 40-nmol bolus followed by saline infusion, and a control group received saline. Intrathecal drug treatment reduced infarct volumes relative to controls by 61%, 31%, and 10% in C(40)C, C(40)S, and C(20)C groups, respectively, but also induced dose-dependent paralysis and elevated mortality. All rats had PID rates similar to the control group prior to treatment, but following treatment secondary phase PIDs were reduced by 47--57% in each drug group compared to controls. Because several animals exhibited PID inhibition but no neuroprotection, there was no significant correlation between these endpoints across groups. However, drug-treated animals that did not exhibit secondary phase PIDs prior to treatment had significantly smaller infarcts and reduced subsequent PID activity than corresponding control rats. Results suggest that post-reperfusion PIDs play a substantial, though still undefined pathogenic role in delayed maturation of cerebral infarction and NMDA receptor-targeted neuroprotection.