Objective <p>Electric phrenic nerve stimulation is employed as a method of artificial ventilation, and computational models are utilized to assist in parameter selection. The majority of models assume isotropic tissue conductivity, although muscle tissue exhibits anisotropic properties. We aim to investigate the influence of anisotropic muscle conductivity on the results of phrenic nerve activation.</p> Methods <p>To calculate the potential distribution, we used an anatomically detailed multi-scale model for non-invasive electrical stimulation in the neck, incorporating realistic muscle fiber orientations. Phrenic nerve activation thresholds were calculated using the McIntyre-Richardson-Grill nerve model. Anisotropy ratios ranging from 1:1 to 1:15 (transversal:longitudinal conductivities) were analyzed at constant corresponding isotropic conductivity. Additional simulations assessed the influence of muscle volume and electrode placement and quantified possible co-activation of other nerves in the neck.</p> Main results <p>Increasing anisotropy ratios resulted in consistently higher phrenic nerve activation thresholds across all axon diameters (up to + 90%). Larger muscle volumes and electrode positions directly over a muscle further elevated the anisotropy effects. Considering anisotropic muscle conductivity increases the number of co-activated nerves.</p> Conclusion <p>High-resolution models incorporating anisotropic conductivity are recommended for research studies on phrenic nerve stimulation.</p> Graphical abstract <p></p>