Results from two retrieval algorithms, "o3_aer" and "o3_mlr", used for SAGE III solar occultation ozone measurements in the stratosphere and upper troposphere are compared. The main differences between these two retrieved (version 3.0) ozone are found at altitudes above 40 km and below 15 km. Compared to correlative measurements, the SAGE II type ozone retrievals (o3_aer) provide better precisions above 40 km and do not induce artificial hemispheric differences in upper stratospheric ozone. The multiple linear regression technique (o3_mlr), however, can yield slightly more accurate ozone (by a few percent) in the lower stratosphere and upper troposphere. By using SAGE III (version 3.0) ozone from both algorithms and in their preferred regions, the agreement between SAGE III and correlative measurements is shown to be ~5% down to 17 km. Below 17 km SAGE III ozone values are systematically higher, by 10% at 13 km, and a small hemispheric difference (a few percent) appears. Compared to SAGE III and HALOE, SAGE II ozone has the best accuracy in the lowest few kilometers of the stratosphere. Estimated precision in SAGE III ozone is about 5% or better between 20 and 40 km and ~10% at 50 km. The precision below 20 km is difficult to evaluate because of limited coincidences between SAGE III and sondes. SAGE III ozone values are systematically slightly larger (2--3%) than those from SAGE II but the profile shapes are remarkably similar for altitudes above 15 km. There is no evidence of any relative drift or time dependent differences between these two instruments for altitudes above 15--20 km.