Analysis of shear‐wave splitting requires records of well‐isolated shear waves uncontaminated by other phases. Because contamination by P‐to‐S or S‐to‐P converted waves occurs easily, we have to choose the epicentral‐distance window carefully to minimize such effects. We here provide several diagrams for each of the direct s, S, ScS, and ScS2 waves, which show how systematically converted waves bias the result of a splitting analysis as a function of epicentral distance and focal depth. The diagrams were constructed using synthetic seismograms calculated up to either 2 or 0.25 Hz for a standard radially symmetric and isotropic Earth model. The data selection based on these diagrams is intended as a first step in a data selection protocol.