Analysis of anisotropy in single-molecule fluorescence experiments using the probability distribution analysis (PDA) method is presented. The theory of anisotropy-PDA is an extension of the PDA theory recently developed for the analysis of Förster resonance energy transfer (FRET) signals [Antonik, M.; et al. J. Phys. Chem. B 2006, 110, 6970]. The PDA method predicts the shape of anisotropy histograms for any given expected ensemble anisotropy, signal intensity distribution, and background. Further improvements of the PDA theory allow one to work with very low photon numbers, i.e., starting from the level of background signal. Analysis of experimental and simulated data shows that PDA has the major advantage to unambiguously distinguish between shot noise broadening and broadening caused by heterogeneities in the sample. Fitting of experimental histograms yields anisotropy values of individual species, which can be directly compared with those measured in ensemble experiments. Excellent agreement between the ensemble data and the results of PDA demonstrates a good absolute accuracy of the PDA method. The precision in determination of mean values depends mainly on the total number of photons, whereas the ability of PDA to detect the presence of heterogeneities strongly depends on the time window length. In its present form PDA can be also applied to computed fluorescence parameters such as FRET efficiency and scatter-corrected fluorescence anisotropy. Extension of the PDA theory to low photon numbers makes it possible to apply PDA to dynamic systems, for which high time resolution is required. In this way PDA is developed as a sensitive tool to detect biomolecular heterogeneities in space and time.