RATIONALE: Functional reinnervation of the transplanted human heart by the autonomic nervous system has not been demonstrated. A lack of autonomic control of the transplanted allograft is reflected by an increased resting heart rate, a sluggish heart rate response to dynamical exercise and a reduced heart rate variability. Recent evidence suggests that a measure of deterministic chaos in the heartbeat interval series (point correlation dimension, PD2i) is superior to the conventional power spectrum or other stochastic measures in detecting changes in the mechanism underlying heartbeat generation. METHODS: The PD2i is based on the presumption that the variability is determined and patterned, whereas the stochastic measures all assume that the variability is around a stationary mean and is noise. The PD2i reconstructs the degrees of freedom (number of independent variables) in the system that generates the time series examined, and does this irrespective of whether the system is stochastic or deterministic and is stationary in time. RESULTS: PD2i was determined for heartbeat intervals (RR, ECG digitized at 1200 Hz; supine posture) of 23 heart transplant recipients (HTR: 9 adults, 14 children; 0.04-7.7 years after transplantation) and 21 healthy control subjects (CTL; 13 adults, 8 children). The PD2i (+/-SD) averaged 5.4 +/- 0.7 for the CTL adults and 5.4 +/- 0.6 for the CTL children. Mean PD2i was reduced after transplantation to 1.1 +/- 0.1 in 6 HTRs recorded within 1 year after surgery; in one HTR recorded 2 weeks after surgery the mean PD2i was 3.7. Between 1 to 2 years PD2i was found increased in 2 of 3 subjects and between 2 to 8 years it was increased in 13 of 13, but not to control levels. In normal hearts the QT subinterval of each heartbeat cycle is associated with inotropy and the RR-QT remainder with chronotropy (i.e., the dyastolic interval during which RR is primarily regulated). To examine more closely the residual and returning heartbeat dynamics of the HTR subjects, these subinterval series were examined during mild exercise (40 to 90 Watts) and its recovery. In recent HTRs, resting QT and RR-QT were moderately reduced and modulated by exercise and recovery, but with an approximate 100 beat latency. In long-term (7-8 years) HTR subjects there was a rapid and larger response to exercise/recovery, but compared to normal the range was smaller and the complexity of the subinterval trajectories in time was simpler. CONCLUSIONS: Recurrence of low-dimensional deterministic dynamics after transplantation suggests recovery of neurocardiac control attributable to 1) reorganization of the viable intrinsic cardiac nervous system, 2) reinnervation by the extrinsic autonomic nervous system, or 3) both.