This study presents a survey of the results obtained by the authors on statistical description of dynamical chaos and the effect of noise on
dynamical regimes. We deal with nearly hyperbolic and nonhyperbolic chaotic
attractors and discuss methods of diagnosing the type of an attractor. We
consider regularities of the relaxation to an invariant probability measure for
different types of attractors. We explore peculiarities of autocorrelation decay and of power spectrum shape and their interconnection with Lyapunov
exponents, instantaneous phase diffusion and the intensity of external noise.
Numeric results are compared with experimental data.