Super-resolution fluorescence microscopy is a promising tool with the potential to strengthen our understanding of living processes. Based on the ability to switch fluorophores on and off in an either deterministic or stochastic manner, the fundamental limit of diffraction can be overcome. The first concepts have been invented 20 years ago, but first discoveries in biology have been reported only recently. As with every new technology, its successful application is challenging. In particular, super-resolution microscopy sets much higher demands on the sample preparation and labeling quality than researchers are used to. In addition, the super-resolution microscopes often require tedious alignment procedures and/or rely on complicated data analysis software. Not surprisingly, commercially available systems are rather expensive and not yet widely spread. This thesis focused on the development and characterization of novel concepts for super-resolution microscopy based on stochastic fluorescence fluctuations in order to extend and/or simplify existing techniques. In particular, we developed a framework for the image-based simultaneous estimation of three-dimensional position and orientation of fluorescent emitters. Unlike many localization-based super-resolution techniques, our approach accounts for the dipole characteristics of the fluorescence emission whilst staying compatible with standard widefield fluorescence microscopes. In a comparative study of two well-established techniques for super-resolution microscopy, namely super-resolution optical fluctuation imaging (SOFI) and stochastic optical reconstruction microscopy (STORM), we determined their characteristics and demonstrated a complementary performance, which suggested a beneficial impact upon combination of both techniques. This has led to a further development of the original SOFI analysis. The resulting balanced SOFI analyzes molecular statistics and linearizes the response with respect to brightness and blinking heterogeneities in the sample, which significantly improves the image contrast and thereby facilitates the access to higher resolutions. We experimentally demonstrated nearly five-fold resolution improvements as compared to diffraction-limited images of fluorescently labeled cells. The super-resolved molecular parameter maps obtained with balanced SOFI, such as the blinking lifetimes, fluctuation amplitudes and label densities, are sensitive to static differences and/or dynamic changes in the chemical microenvironment of the fluorophores and can thus report functional information that has not been exploited before, for instance, the local pH or the concentration of reducing and oxidizing agents. Finally, by using cross-cumulation between multiple depth and/or color channels, we demonstrated whole-cell three-dimensional super-resolution microscopy with a normal widefield illumination and without mechanical scanning as well as spectral unmixing of multiple fluorophores with overlapping emission spectra.