Since atomic force microscopy (AFM) images are a composite of probe and sample geometry, accurate size determinations are problematic. A relatively straightforward mathematical procedure for determining tip radius of curvature (R-T) for an asymmetrical tip was recently developed by Garcia et al. (Probe Microsc. 1997, 1, 107). This study represents an experimental test of that procedure for both silica (similar to 150 nm) and polystyrene (similar to 50 nm) nanospheres. The procedure can be summarized by two steps : (1) tip characterization assuming that the observed AFM height is a true measure of a spherical particle＇s diameter and (2) use of the tip shape to extract a calculated width. To ensure that AFM heights were equivalent to the true width, a direct comparison of individual particle sizes determined by transmission electron microscopy (TEM) and AFM was conducted. Heights measured from AFM images of polystyrene nanospheres differed, on average, less than 5% from widths measured by TEM. The quality of R-T values was therefore evaluated by the magnitude of relative error in calculated particle widths with respect to true widths. For the tip used in this study a calculated R-T of 13 nm resulted in excellent calculated widths for both polystyrene and silica spheres. While spherical particles whose diameter is less than R-T (such as 5-nm Au colloids) can be used to characterize the tip apex, larger diameter spheres are required to fully characterize the tip. However, spheres much larger than R-T predominantly interact with the walls of the tip and therefore yield artificially high R-T values. On the basis of our analysis of the procedure developed by Garcia et al., the best sphere size for full characterization of the tip (apex and walls) is one in which both portions of the tip interact with the sphere to similar extents (approximately : R-T less than or equal to R-P less than or equal to R-T, where R-P is the particle radius).