Several improvements for a previously described semiautomated system for spin-wave instability threshold measurement in microwave ferrites are described. The effects of sample size, sample position in the microwave cavity, and method of data interpretation on the threshold field hcrit are also reported. The data were obtained by the parallel pumping technique at 8.92 GHz. Sphere-shaped samples of polycrystalline yttrium iron garnet were used for the measurements. The results show that the initial, ‘‘exact’’ instability onset, the ‘‘extrapolated’’ onset from the loss increase above threshold, and the ‘‘visual’’ onset estimate from the above threshold response can vary by more than 0.5 Oe for hcrit values of 3 Oe or so. Large samples, above 1 mm in diameter, lead to an overestimation of hcrit due to errors in the microwave cavity perturbation theory analysis which forms the basis of the measurement. The amount of overestimation is about 0.1 Oe for a diameter of 2 mm and 0.2 Oe at 3 mm. Sample position relative to the cavity end wall does not appear to be critical. Imaging effects are small for hcrit measurements in the YIG-like materials of interest here.