Purpose In this study, we evaluated the effects of perfusion of the skeletal muscle on diffusion tensor imaging (DTI) and diffusional kurtosis imaging (DKI) parameters and their reproducibility. Methods Diffusion tensor imaging and DKI models, with and without intravoxel incoherent motion (IVIM) correction, were applied to simulated data at different physiological conditions and signal‐to‐noise ratio levels. Next, the same models were applied to data of the right calf of five healthy volunteers, acquired twice at 3 telsa. For six muscles, we evaluated the correlation of the perfusion signal fraction, with parameters derived from DTI and DKI, and performed repeatability analysis with and without IVIM correction. Additionally, the IVIM correction was compared to a multishell acquisition approach that minimizes perfusion effects on DTI estimates. Results Simulations and acquired data showed that DTI and DKI estimates were biased proportionally to the perfusion signal fraction, and that IVIM correction was needed for accurate estimation of the DTI and DKI parameters. However, taking perfusion into account did not improve repeatability. Conclusion Blood perfusion has an effect on DTI and DKI estimations, but it can be minimized with IVIM correction or multishell acquisition strategies. Magn Reson Med 78:233–246, 2017. © 2016 International Society for Magnetic Resonance in Medicine