Résumé : Objective: We previously found that using a bimodal oddball design with synchronized pairs of audio-visual stimuli increased the sensitivity of the P300 wave to subclinical differences in depression. We wondered: (1) whether these P300 modulations were due to facilitated discrimination of deviant vs. frequent stimuli and (2) whether congruency of the stimuli was necessary to produce this effect. Method: Two groups of participants (students displaying anxious-depressive tendencies as the subclinical group and healthy students as the control group) had to detect deviant stimuli among frequent stimuli in oddball tasks. Experiment 1 involved either single faces (Single condition) or two similar faces presented side-by-side (Double condition). Experiment 2 involved audio (A), visual (V), bimodal congruent (AVC), and bimodal incongruent (AVI) oddball tasks. Results: While the easiest Double condition produced the fastest reaction times and shortest P300 latencies, there were no significant differences in P300 modulation between the groups in Experiment 1. In Experiment 2, P300 amplitudes of the control group were higher in response to AVC tasks than in response to unimodal (A or V) or AVI tasks. Conclusions: The increased sensitivity of the P300 wave to subclinical differences that we observed here is not related to an enhanced discrimination effect. However, during bimodal tasks, stimulus congruency is necessary for the appearance of the increased P300 sensitivity. Significance: The impaired "bimodal congruence effect" in patients with subclinical depression/anxiety suggests these patients have altered integrative processes, which has potential implications for cognitive therapy.