Audiovisual early integration was investigated in humans using behavioral and electrophysiological measures in the attending one-modality (visual/auditory) condition to determine whether audiovisual early integration occurs when one modality is attended. Visual, auditory, and audiovisual stimuli were presented randomly with equal probability. The subjects were instructed to press a button in response to an infrequent target stimulus (20%) in the attend-designated modality. Reaction times (RTs) and accuracy were recorded and were compared using paired t-tests. The event-related potentials (ERPs) to bimodal stimuli (AV) were compared with the sum of the ERPs to unimodal auditory (A) and visual (V) stimuli. The early integration process was expressed as the difference waveform between AV and (A+V). Audiovisual early integration (P100) of the bimodal audiovisual stimuli was found in the scalp topographies in the attending auditory condition; moreover, the RTs to the bimodal audiovisual target stimuli were faster than the RTs to the unimodal auditory target stimuli. By contrast, in the attending visual condition, audiovisual early integration (P100) of the bimodal audiovisual stimuli was not found in the scalp topographies, and no significant RT difference was observed between the unimodal visual and bimodal audiovisual target stimuli.