A commercially available magnetic therapy system, designed for clinical application as well as for private use without medical supervision, was examined with respect to its potential for causing electromagnetic interference with implantable pacemakers (PMs) and automatic implantable cardioverter defibrillators (AICDs). A sample of 15 PMs and 5 AICDs were experimentally investigated. Each of the implants was realistically positioned in a homogeneous, electrically passive torso phantom and exposed to the magnetic fields of the system's applicators (whole body mat, cushion, and bar applicator). The detection thresholds of the implants were programmed to maximum sensitivity and both unipolar as well as bipolar electrode configurations were considered. The evaluation of possible interferences was derived from the internal event storages and pacing statistics recorded by the implants during exposure. Any "heart activity" recorded by the implants during exposure was interpreted as a potential interference, because the implant obviously misinterpreted the external interference signal as a physiological signal. Only cases without any recorded "heart activity" and with nominal pacing rates (as expected from the program parameter settings) of the implants were rated as "interference-free." Exposure to the whole body mat (peak magnetic induction up to 265 microT) did not show an influence on PMs and AICD in any case. The cushion applicator at the highest field intensity (peak magnetic induction up to 360 microT) led to atrial sensing defects in four PM models with unipolar electrode configuration. Under bipolar electrode configuration no disturbances occurred. The bar applicator led to sensing problems and consecutively reduced pacing rates in all tested PM models under unipolar electrode configuration and maximum field intensity (peak magnetic induction up to 980 microT). Bipolar electrode configuration resolved the problem. The investigated AICDs did not show malfunctions under any investigated condition. In conclusion, the examined PEMF therapy system did not interfere with the investigated implantable cardiac devices with bipolar electrode configuration. However, unipolar electrode configuration in pacemakers seems to be potentially hazardous during application of the examined PEMF therapy system.