OBJECTIVE We investigated the influence of the number of stimuli on signal-to-noise (S/N) ratio of CSERP. METHODS CSERP from 20 normosmic subjects were obtained in response to stimulation with two olfactory (H(2)S and PEA) and a trigeminal (CO(2)) stimulant. For each of these odors, 160 stimuli were delivered into the right nostril (duration 200ms, mean ISI 30s) using a constant-flow, air-dilution olfactometer. For each EEG recording site (Fz, Cz, Pz, C3, C4), peak-to-peak amplitude N1P2 and noise amplitude levels were determined. Subsequently, S/N ratios were calculated. RESULTS The S/N ratios for olfactory ERP generally improved for H(2)S and PEA. For responses to PEA, S/N ratios increased significantly up to 80 averages (S/N ratio=5.6). The number of stimuli for an optimal S/N ratio for trigeminal ERP was slightly lower, i.e. 60 averages (S/N ratio=7.9). CONCLUSIONS S/N N1P2 ratios in olfactory and trigeminal ERP significantly improve with an increasing number of responses averaged under these experimental conditions. This is mainly due to a reduction of noise level. Applying more stimuli has little additional effect on S/N ratio due to a concomitant decrease in signal amplitude. SIGNIFICANCE An optimal S/N ratio is essential when recording CSERP in neurodegenerative disorders, where responses may be of low amplitude, and for medico-legal purposes.