In this study, we investigate correlation properties of fluctuations in heart interbeat (RR) time series in a broad range of physiological and pathological conditions. Using detrended fluctuation analysis (DFA) method we determined short-term (alpha 1) and long-term (alpha 2) scaling exponent. In addition, we calculated standard deviation of RR intervals (SDRR) as the simplest variability measure. We found that the difference between alpha 1 and alpha 2 is related to RR interval length. At the shortest RR intervals, which correspond to extreme physiological and pathological conditions, we found the highest reduction of variability and the biggest difference between scaling exponents. In this case, DFA reveals a white noise over short scales (alpha 1 about 0.5) and strongly correlated noise over large scales (alpha 2 about 1.5). With an increase in RR interval, accompanied by increased variability (increase in parasympathetic control), the difference between alpha 1 and alpha 2 decreases. The difference between scaling exponents disappeared in a state of efficient autonomic control. We suggest that the complexity in heart rhythm is achieved through coupling between intrinsically controlled heart rhythm and autonomic control, and that the model of stochastic resonance mechanism could be applied to this system.