This study presents a preliminary analysis of the relationship between electroencephalographic (EEG) and electrocorticographic (ECoG) event-related potentials (ERPs) recorded from from a single patient using a brain-computer interface (BCI) speller. The patient had medically intractable epilepsy and underwent temporary placement of an intracranial ECoG grid electrode array to localize seizure foci. The patient performed one experimental session using the BCI spelling paradigm controlled by scalp-recorded EEG prior to the ECoG grid implantation, and one identical session controlled by ECoG after the grid implantation. The patient was able to achieve near perfect spelling accuracy using EEG and ECoG. An offline analysis of the average ERPs was performed to assess how accurately the average EEG ERPs could be predicted from the ECoG data. The preliminary results indicate that EEG ERPs can be accurately estimated from proximal asynchronous ECoG data using simple linear spatial models.