A method is disclosed for recording and evaluating PET data recorded at the same time as magnetic resonance data using a combined magnetic resonance/PET apparatus, wherein in the context of a pulse sequence for recording the magnetic resonance data a magnetic resonance recording facility, including at least one gradient coil and at least one high-frequency coil, is activated. In at least one embodiment, the method includes recording the PET data and assigning the recorded PET data after the recording time point to at least two data groups assigned to a predetermined operating state of the magnetic resonance recording facility; for each data group, determining a measure of similarity of the PET data contained therein to the PET data as a whole and/or to PET data of at least one further data group, whereby if the measure of similarity is below a threshold value, the PET data of the data group is rejected and/or further evaluated separately.