To compare different electroencephalography (EEG)-based regressors and their ability to predict the simultaneously recorded blood oxygenation level dependent response during blocked visual stimulation, simultaneous EEG-functional magnetic resonance imaging in 10 healthy volunteers was performed. The performance of different single-trial EEG regressors was compared in terms of predicting the measured blood oxygenation level dependent response. The EEG-based regressors were the amplitude and latency of the primary positive (P1) and negative (N2) peaks of the visual evoked potential, the combined P1-N2 amplitude, and the alpha power. Apart from peak latencies, all regressors showed significant positive or negative correlation with the blood oxygenation level dependent response in visual cortex. In addition, several EEG-based regressors were found to predict blood oxygenation level dependent variations in different occipital and extraoccipital cortical areas not explained by the boxcar regressor. The results suggest that the P1-N2 regressor is the best EEG-based regressor to model the visual paradigm, but when looking for additional effects like habituation or attention modulation that cannot be modeled by the boxcar regressor, it is better to include regressors based on individual peaks or alpha power.