We report results of 13-cm-wavelength radar observations of 1685 Toro conducted in July 1980 at the Arecibo Observatory. Our data yield detections of radar echoes in the same sense of circular polarization as transmitted (i.e., the SC sense) as well as in the opposite (OC) sense. The radar spectra reveal correlated, approximately twofold variations in radar cross section and spectral bandwidth as functions of rotational phase, with two maxima and two minima per rotation cycle. Our estimate of the ratio of SC to OC echo power, 0.18 +/- 0.04, suggests that most, but certainly not all, of the backscattering is due to single reflections from surface elements that are fairly smooth at decimeter scales. However, the total absence of the sharply peaked spectral signature of quasi-specular scattering requires substantial roughness at some much larger scales(s). When combined with photopolarimetric results of Dunlap et al. (1973), our observations provide constraints on Toro's size, shape, and surface properties. The maximum distance of any part of Toro's surface from the spin axis is between 2.4 and 3.4 km. The ratio of Toro's longest to shortest equatorial widths is between 1.4 and 2.2. Modeling Toro as a homogeneously scattering ellipsoid yields weighted-least-squares estimates for the lengths of the equatorial semiaxes: a = 2.60 +/- .0.10 km and b = 1.68 +/- 0.17 km, and a nearly Lambertian scattering law. The magnitude of the post-fit residuals suggests systematic departures of Toro from this simplified model, including a possible surface feature with enhanced radar reflectivity and depolarization. The length, 2c of the rotation axis cannot be estimated from our data, as it is highly anticorrelated with the radar reflectivity. If we assume c <= b, our ellipsoid provides (pi)(ab) = 13.7 +/- 1.8 km^2 as an estimate for an upper limit on Toro's projected area. Using this result, we derive approximate lower limits on Toro's 13-cm and B-filter geometric albedoes (0.04 and 0.23, respectively) which constrain the composition and porosity of Toro's surface. Our lower bound on Toro's B-filter albedo is substantially higher than the value (0.14) reported by Dunlap et al.