Non-invasive monitoring of serum potassium ([K + ]) and calcium ([Ca 2+ ]) concentration can help to prevent arrhythmia in kidney patients. Current electrocardiogram (ECG) markers, including the T wave width (T w) and its time-warped temporal morphological variability (d U w), correlate significantly with [K + ] and [Ca 2+ ] but these relations vary strongly between patients. We hypothesized that inter-individual differences in cell type distribution across the ventricular wall can explain this variability. We computed T w and d U w in simulated ECGs from a human heart-torso model at different proportions of endo-, mid-, and epicardial cells, while varying [K + ] (3 to 6.2 mM) and [Ca 2+ ] (1.4 to 3.2 mM). Electrical activity was simulated with a reaction-diffusion model with modified Ten Tusscher-Panfilov dynamics. Results were compared to data from 29 patients. T w and d U w correlated strongly with [K + ] (absolute median Pearson coefficient r = 0.70 to 0.93) and [Ca 2+ ] (r = 0.69 to 0.86) in simulations and in patients. Different cell type distributions reproduced inter-patient variability, with the same sign and magnitude of r. In conclusion, the inter-patient variability in the relation between serum electrolytes and their ECG markers can indeed be explained by inter-individual differences in cell type distribution across the ventricular wall.