A novel approach to assessment of individual 24-hour heart rate variability (HRV) has been suggested. This approach allows to compare sequences of RR intervals with knowingly nonstationary rhythm and to obtain valuable information on direction (increase or decrease) of HRV changes. The latter was not achievable by existing techniques. Main feature of the novel approach is consideration of dependence of sinus arrhythmia on heart rate on a given time interval (rhythmogram fragment). This became possible due to use of implementation of a new characteristic of 24-hour HRV - mean weighted variation of rhythmogram (MWVR). This parameter repesented a generalization of a previously introduced variation of short rhythmogram fragment (VSRF) and was formed in the following way. Rhythmogram was divided into short fragments and within each fragment VSRF was calculated. Then average variations were averaged for different ranges of HR changes. Each averaged variation was multiplied by weight coefficient which depended on corresponding range of HR changes, and by portion (expressed in percent) of overall number of short rhythmogram fragments belonging to this range. MWVR was determined as sum of resulting multiples. Analysis of rhythmograms from 58 healthy persons showed that MWVR declined with age. In patients with heart failure (n=80) mean MWVR was lower than in healthy persons and decreased with increase of NYHA class of heart failure. Eight patients were restudied while they were classified to a different heart failure class. Transition to a higher class was associated with lowering of MWVR and vice versa. Thus MWVR adequately reflected dynamics of functional state of an individual patient.