Microsecond rotational motions of nitroxide spin labels are measured primarily with saturation transfer electron paramagnetic resonance (ST-EPR). In the present study we have used model system experiments to quantitatively evaluate different ST-EPR spectral parameters, both in-phase and out-of-phase, with an emphasis on techniques for suppressing the interference from weakly immobilized probes. Analyses of both systematic and random errors show that maximum sensitivity to small changes in correlation time and minimum ambiguity of interpretation are best achieved by combining measurements of both spectral line-shape, i.e., the ratio of line-heights, and spectral intensity, i.e., the absolute amplitude of either a position within a spectrum or a spectral integral. Errors in the measurement of correlation times for the two types of parameters tend to be complementary. Integrated intensity parameters are particularly useful in measuring microsecond probe motions in the presence of weakly immobilized components. We confirm that integrated intensity parameters are sometimes effective in rejecting signals from weakly immobilized probes, but the effectiveness of this rejection is more limited than previously supposed and depends on the type of parameter being measured. We describe procedures for evaluating and minimizing errors due to weakly immobilized probes, emphasizing the advantages of a new kind of intensity parameter obtained from integrated in-phase spectra. We provide detailed descriptions of experimental procedures, along with calibration plots of the most useful spectral parameters vs. rotational correlation time, which should make it possible for workers in other laboratories, using different instruments and sample geometries, to reproduce spectra quantitatively and to make accurate correlation time measurements.