During our daily lives we regularly imagine natural scenes, whether we are reading our favorite novel or trying to spatially navigate through the environment. Previous EEG/MEG research has suggested that cortical alpha activity plays a critical role in mediating visual imagery. However, our current understanding of what information is encoded in imagery-related alpha activity is still limited. To investigate this further, we conducted a study in which we aimed to decode the properties of imagined scenes from rhythmic cortical activity patterns, hypothesizing that these properties would be represented in the alpha activity. Participants were presented detailed three-sentence descriptions of individual scenes which varied in four properties: spatial expanse, naturalness, clutter level and luminance. Participants were instructed to read the description and, following a visual cue, vividly imagine the scene. Meanwhile their neural activity was recorded with EEG. Using multivariate decoding analyses on EEG power patterns at specific neural frequencies, we found that all four investigated scene properties were discriminable exclusively in the alpha frequency range. Cross-decoding between the imagery data and data from an additional perception task, in which participants viewed images that matched the scene descriptions, indicated that scene representations in the alpha band are shared between imagery and late perception. Our results show that the properties of imagined scenes are represented in cortical activity in the alpha range, suggesting that alpha activity mediates the top-down activation of scene-related perceptual contents during visual imagery.