Structured illumination microscopy (SIM) is a powerful technique for obtaining super-resolved fluorescence maps of samples, but it is very sensitive to aberrations or misalignments affecting the excitation patterns. Here, we present a reconstruction algorithm that is able to process SIM data even if the illuminations are strongly distorted. The approach is an extension of the recent blind-SIM technique, which reconstructs simultaneously the sample and the excitation patterns without a priori information on the latter. Our algorithm was checked on synthetic and experimental data using distorted and nondistorted illuminations. The reconstructions were similar to that obtained by up-to-date SIM methods when the illuminations were periodic and remained artifact-free when the illuminations were strongly distorted.