Being able to predict self-generated sensory consequences is an important feature of normal brain functioning. In the auditory domain, self-generated sounds lead to smaller brain responses (e.g., auditory evoked responses) compared to externally generated sounds, which is usually referred to as the sensory attenuation effect. Here we investigated the role of brain oscillations underlying this effect. With magnetoencephalography, we show that self-generated sounds are associated with increased pre-stimulus alpha power and decreased post-stimulus gamma power and alpha/beta phase locking in auditory cortex. All these oscillatory changes are correlated with changes in evoked responses, suggesting a tight link between these oscillatory events and sensory attenuation. Furthermore, the pre- and post- oscillatory changes correlate with each other across participants, supporting the idea that they constitute a neural information processing sequence for self-generated sounds. In line with findings of alpha oscillations reflecting feedback and gamma oscillations feedforward processes and models of predictive coding, we suggest that pre-stimulus alpha power represent prediction and post-stimulus gamma power represent prediction error, which is further processed with post-stimulus alpha/beta phase resetting. The correlation between these oscillatory events is further validated with cross-trial analysis, which provides additional support for the proposed information processing sequence that might reflect a general mechanism for the prediction of self-generated sensory input.