Choice responses are faster when target position and response side correspond than when they do not, even if target position is response-irrelevant. This Simon effect has also been observed in case of multi-item arrays. Generally, it is assumed that an automatically generated spatial response code is responsible for the effect. The referential-coding account assumes that this code is directly related to the target, although the moment of production of the code is not fully clear. The attention-shift account assumes that the code is directly related to the direction of the most recent attentional shift. An experiment was performed in which left or right target locations were indicated by arrows occurring before (precue), simultaneous with (simcue), or after (postcue) six-element arrays. Overt responses and EEG potentials were recorded. The Simon effect was present in all conditions, and decreased when responses were slower. No relation was found between amplitude of posterior lateralized components and the magnitude of the Simon effect. A posterior contralateral negativity was also found after presenting the arrays in the precue condition, which might reflect the reorienting of attention toward the target position. The results are more favorable to the referential-coding account although this account becomes very similar to the attention-shift hypothesis as the moment of formation of the spatial response code is related to effective target onset rather than to stimulus onset.