This study evaluates the problems and the evolution of cardiac stimulation in infants (aged < 3.5 years) by comparing the endocavity and epicardial routes in a retrospective series of 37 patients. Thirty seven patients aged 1.2 +/- 0.9 years treated with epicardial (n = 19) or endocavity (n = 18) stimulation were followed for 10.9 +/- 6.4 years (0.75-24). The 2 patient groups did not differ in age or weight. Four patients were lost to follow up, and 1 died. The functional duration of the first stimulator was not significantly different if the initial approach was epicardial or endocavity. The endocavity probes were introduced by venous denudation in 15 cases and by subclavian puncture in 3 cases. Fourteen of the 19 children fitted by the epicardial route went on to endocavity stimulation, of which 10 were at the first replacement. None of the 18 patients fitted by the endocavity route went on to epicardial stimulation. Out of 11 endocavity probe replacements and 9 atrialisations, the homolateral venous approach was always possible except in 2 cases. In conclusion, the results for the epicardial and endocavity routes are comparable. For technical reasons (calibre of the veins, size of the stimulator) it would appear reasonable if the endocavity route was used, making do initially with a mono chamber stimulation. The advances in the epicardial electrodes abolishes the major handicap (threshold elevation) of this approach which can be advocated when double chamber stimulation seems preferable.