Research on conscious somatosensory perception has mainly relied on electrical stimulation. Here, we directly compare the event-related potentials evoked by tactile vs. electrical peri-threshold stimulation, using a matching task to control for post-perceptual processes. EEG was recorded while participants received either electrical pulses to the left median nerve (n=24) or tactile stimulation at the left index finger (n=25). To control for confounding processes, participants did not directly report their experience, but indicated via saccades whether they experienced a match or a mismatch between their tactile experience and a simultaneously presented visual cue. Because conscious perception is highly collinear with stimulus intensity and detection probability, we used Bayesian model selection to determine which of a set of similar models best explains post-stimulus ERP activity, both within and across datasets. We find that early potentials (P50 and P100) were attenuated for undetected relative to detected trials, only for tactile but not for electrical stimuli. Moreover, an early (~90 ms) negative component was present only for detected tactile stimulation. While the early and mid-latency components were driven mostly by detection and detection probability during tactile stimulation and by stimulus intensity during electrical stimulation, late (>300 ms) components were mostly driven by detection during tactile stimulation and by detection probability during electrical stimulation. Our findings suggest that post-stimulus processing is dependent on the type of stimulation used, and that early ERP amplitudes may already index tactile detection. This contrasts with previous research using electrical stimulation that had indicated mere scaling with stimulus intensity.