A new numerical analysis method for experimental single-pair fluorescence resonance energy transfer (sp-FRET) data is proposed. In this method, every single data point was plotted in a style of a cumulative distribution function and dedicated to curve-fitting analysis, so that the analysis does not depend on bin size. A series of numerical simulations showed that this analysis has a more,efficient and accurate resolvability of components than a fitting method based on Gaussian functions to a histogram plot. A simulated data based on experimental FRET distributions were also used to discuss the fitting errors of this method. The proposed method was applied to sp-FRET experiments of doubly dye-labeled double-strand DNA with a short sequence. Mixtures of up to three species were analyzed, and the contributions up to four subpopulations were successfully resolved.