Methods and apparatuses for detecting sleep apnea by analyzing characteristic physiological oscillations of the heart rate variability (HRV). Starting from recorded ECG data of the patient, for example, as a long-term sequence of the changing RR intervals, the heart rate variability is examined using autocorrelation calculations for the occurrence of rhythmic oscillations of various frequencies. If oscillations typical for apnea occur having very long period durations in the range of 20 to 80 seconds, these are detected as a maximum of the autocorrelation function. If a pathological sleep apnea accordingly exists, individual apnea events may be identified by prompt analysis of short recorded RR sequences, e.g., in the minute interval.