Observations using single-molecule localization microscopy have led to the belief that the majority of tested membrane proteins are organized in clusters at sizes below the diffraction limit. These nanoclusters are thought to play an important role in cellular signaling. However, concerns about the existence of nanoclusters have been fueled by the notion that virtually all fluorescent probes show complex blinking behavior including long-lived dark states. This results in localization clusters due to the repeated observation of single molecules. Existing post-processing approaches commonly struggle to reliably distinguish real molecular clustering from such blinking artifacts. In this thesis, a novel analytical method is presented that uses information from two-color STORM experiments for reliably detecting molecular clustering while overcoming the erroneous detection of clustering due to fluorophore blinking. Targeting the same protein species with differently labeled antibodies allows for the calculation of distance distributions between localizations from both color channels. Molecular clusters exhibit a characteristic bias towards shorter distances. Applying toroidal shifts to the data breaks possible correlations between the two color channels, thus providing surrogates for realizations of the null hypothesis of randomly distributed molecules. This allows for statistical signiffcance tests without the necessity of additional calibration. This work evaluates the limits of the method with Monte Carlo simulations and experiments on clustered and randomly distributed membrane proteins.