We present a formal error analysis and characterization of the microwave measurements made during the Stratospheric Ozone Intercomparison Campaign (STOIC). The most important error sources are found to be determination of the tropospheric opacity, the pressure-broadening coefficient of the observed line, and systematic variations in instrument response as a function of frequency (''baseline''). Net precision is 4--6% between 55 and 0.2 mbar, while accuracy is 6--10%. Resolution is 8--10 km below 3 mbar and increases to 17 km at 0.2 mbar. We show the ''blind'' microwave measurements from STOIC and make limited comparisons to other measurements. We use the averaging kernels of the microwave measurement to eliminate resolution and a priori effects from a comparison to SAGE II. The STOIC results and comparisons are broadly consistent with the formal analysis. ¿ American Geophysical Union 1995