Although the apnea/hypopnea index is the most widely used measure of breathing pattern abnormality during sleep, this index gives no information about the strength of the oscillation in the breathing pattern, its periodicity or its regularity. Such information may be required in research studies involving breathing patterns and how they are affected by interventions. We are exploring spectral analytic methods to determine two normalized indices, the periodicity index and the modified modulation index, to examine periodic breathing for all-night sleep studies. These methods are automatic and require no user interaction. Data were obtained from 11 heart failure patients who slept for a total of 21 nights in the sleep laboratory. Because individual patients had a marked regularity of their Cheyne-Stokes respiration during sleep, one would expect an extremely high correlation between the traditional measures of breathing pattern abnormality and these spectral analytic techniques. Indeed we found that there was an extremely high correlation between the periodicity index and the modulation index and the traditional measures of apnea/hypopnea index and the proportion of the night with periodic breathing (p less than 0.02 in all cases). When the breathing pattern was irregular but still with many apneas there was a discrepancy between the apnea index and the indices of periodicity. These techniques are still preliminary and future studies will determine their limitations in other patient populations and where the pattern is unstable.