BigP3BCI Study G is a P300-based brain-computer interface dataset comprising EEG recordings from 20 healthy subjects performing a 9x8 checkerboard visual speller task. This derivative dataset is part of the larger BigP3BCI collection, the largest public P300 BCI dataset with ~267 subjects across 20 studies. The data were acquired at 256 Hz using 16-channel EEG with a g.USBamp amplifier and include target and non-target event classifications suitable for machine learning applications.