We performed experiments and computer modeling of heating of a cardiovascular stent and a straight, thin wire by RF fields in a 1.5 T MRI birdcage coil at 64 MHz. We used ASTM F2182-02a standard and normalized results to 4 W/kg whole body average. We used a rectangular saline-gel filled phantom and a coiled, double stent (Intracoil by ev3 Inc) 11 cm long. The stent had thin electrical insulation except for bare ends (simulating drug eluting coating). The stent and phantom were placed close to the wall of the RF Coil and had approximately 0.5 degrees C initial temperature rise at the ends (local SAR = 320 W/kg). We exposed a wire (24.1 cm, 0.5 mm diameter) with 0.5 mm insulation and saw an 8.6 degrees C temperature rise (local SAR = 5,680 W/kg) at the bare ends. All heating was within 1 mm3 of the ends, so the position of our fiber optic temperature probe was critical for repeatability. Our computational study used finite difference time domain software with a thermodynamics solver. We modeled a coiled bare-wire stent as a spiral with a rectangular cross section and found a maximum increase of 0.05 degrees C induced at the tips for plane wave exposures. A maximum local SAR of up to 200 W/kg occurred in a volume of only 8 x 10(-3) mm. We developed improved computational exposure sources-- optimized birdcage coils and quasi-MRI fields that may eliminate the need to model an RF coil. We learned that local (point) SAR (initial linear temperature rise) is the most reliable indicator of the maximum heating of an implant. Local SAR depends greatly on implant length, insulation and shape, and position in the MRI coil. Accurate heating must be measured with sensors or software having millimeter resolution. Many commercially available fiber optic temperature probes do meet this requirement.