Commercially available superparamagnetic nanospheres are commonly used in a wide range of biological applications, particularly in magnetically assisted separations. A new and potentially significant technology involves the use of these particles as labels in magnetoresistive assay applications. In these assays, magnetic bead labels are used like fluorescent labels except that the beads are excited and detected with magnetic fields rather than with photons. A major advantage of this technique is that the means for excitation and detection are easily integrable on a silicon circuit. A preliminary study of this technique demonstrated its basic feasibility, and projected a sensitivity of better than 10(-12) molar [Baselt ct al., Biosensors Bioelectronic 13, 731 (1998)]. In this article we examine the theoretical signal to noise ratio of this type of assay for the special case of a single magnetic bead being detected by a single giant magnetoresistive (GMR) detector. Assuming experimentally observed and reasonable parameters for the magnetic label and the sensitivity of the GMR detector, the signal to noise ratio is calculated to be greater than 5000:1 for detection of a single 1 mu m diameter magnetic microsphere immobilized on the surface of a 1 mu m x 1 mu m GMR sensor. Based on this large signal to noise ratio, the detection format should be applicable to more complicated assays where linear quantification is required or to assays requiring significantly smaller beads. Detection of microsphere labels approaching 10 nm may be possible upon further technological advances.