Extracellular field potentials and [K⁺]_o were recorded in slices of human epileptogenic neocortex maintained in vitro during perfusion with Mg²⁺‐free artificial cerebrospinal fluid (ACSF). The human neocortex was obtained during neurosurgical procedures for the relief of seizures that were resistant to medical treatment. Spontaneous epileptiform activity and episodes of spreading depression appeared within 1.5 to 2 hours of perfusion with Mg²⁺‐free ACSF. The epileptiform discharges consisted of negative field potential shifts (amplitude, 0.8–10 mV) that lasted 2.5 to 80 seconds and recurred at intervals ranging between 4 and 160 seconds. Both duration and frequency occurrence of epileptiform events were not significantly different when measured in slices obtained from spiking tissue compared with those gathered from nonspiking neocortical areas. Transient increases in [K⁺]_o of up to 10.5 mM were associated with each epileptiform discharge; these changes were maximal and fastest in the middle neocortical layers. Spreading depression episodes were characterized by 20 to 30‐mV negative shifts that lasted up to 200 seconds and were accompanied by increases in [K⁺]_o of approximately 100 mM. Epileptiform discharges and spreading depressions did not occur during perfusion with Mg²⁺‐free ACSF that contained either competitive or noncompetitive antagonists of the N‐methyl‐D‐aspartate (NMDA) receptor subtype. In contrast, pharmacological blockade of non‐NMDA receptors did not influence the epileptiform activity observed in Mg²⁺‐free ACSF. These findings demonstrate that decreasing [Mg²⁺]_o leads to the appearance of both spontaneous epileptiform discharges and spreading depression in the human epileptogenic neocortex. Both activities are solely dependent on NMDA‐activated conductances, suggesting that removal of the voltage‐dependent gating effect exerted by extracellular Mg²⁺ on the NMDA ionophore is a condition sufficient for neurons in the human neocortex to generate spontaneous, synchronous epileptiform activity. We propose that this experimental paradigm might represent a model suitable for analyzing the cellular mechanisms underlying human neocortical epileptogenesis in vitro as well as for preliminary screening of antiepileptic drugs.