The "EGM width criterion" is a discrimination algorithm that was available in the last generation ICDs. It improved ventricular tachycardia detection by withholding inappropriate therapy deliveries in the presence of narrow QRS tachycardias. The accuracy of the algorithm depends on the optimal settings of the intracardiac EGM source, the "slew thresholds," and the "width threshold." The possible dependence of these parameters on body position may affect the detection efficacy. Whether these effects can be minimized by a proper choice of the electrode configuration used for signal analysis is still to be investigated. This study aimed to evaluate the stability of the slew threshold and width threshold obtained in the supine and orthostatic positions detected by the tip-to-ventricular coil and can-to-ventricular coil electrode configurations. Their time dependence was also evaluated at the 6-month follow-up. Fifty-eight patients who were recipients of an ICD (model Medtronic 7223cx and 7227cx) were included in the study. Changing from supine to orthostatic position caused a marked variation of slew and width thresholds (21.0 +/- 13.9 V/s and 10.1 +/- 9.6 ms, respectively) in 36% of patients with tip-to-ventricular coil and in 44% of patients with can-to-defibrillating coil (the mean slew threshold variation was in this case 17.6 +/- 15.8 V/s, while the mean width threshold variation was 18.8 +/- 21.0 ms). Width threshold variation was statistically significant (P < 0.02) with the latter electrode configuration. Slew thresholds settings changed between the 1- and 6-month follow-ups in the 75% of patients with can-to-defibrillating coil configuration and in 50% with tip-to-defibrillating coil. These time related variations were significantly larger with the tip-to-defibrillating coil configuration (P < 0.01). In conclusion, EGM width parameters may change between supine and orthostatic position and over time with tip-to-defibrillating coil configuration and can-to-defibrillating coil configuration. The former configuration was less sensitive to body position changes, but more sensitive to time related variations. These findings may be useful for optimal programming of the EGM width criterion, but if parameter programming based on these results can improve the discrimination specificity still needs to be investigated.