In recent years continuous flash suppression (CFS) has become a popular “blinding” technique for the investigation of nonconscious affective processing since it elicits potent and long-lasting suppression of conscious visual perception. While the majority of studies provides some positive evidence for nonconscious affective processing, there are also studies reporting their absence. Several methodological variations may give rise to this discrepancy: with respect to the experimental paradigm these variations pertain to the likelihood of residual stimulus visibility on the level of individual participants and single trials. Concerning the statistical analysis they relate to the procedures applied to assess whether detection performance is at chance level and whether the outcome measure does or does not depend on the affective stimulus category. In the present study we determined individual eye dominance and individual stimulus contrast in pretests, measured objective and subjective awareness online and applied Bayesian statistics to estimate the likelihood for the null hypothesis. Event-related potentials (ERPs) were measured while participants were subjected to fearful, happy, and neutral faces in a conscious as well as in a nonconscious CFS condition. In the conscious condition, expected emotion effects were observed in the ERP components N170/EPN and LPP. However, despite high statistical power, no effects of emotional expression were found in the nonconscious condition and the absence of nonconscious affective processing under the tested conditions was substantially more likely than its presence. We discuss whether CFS disrupts affective processing completely if thoroughly applied or whether positive and negative findings should be integrated under a two-threshold framework of nonconscious processing.