Vertical ozone profiles measured with balloon sondes at Uccle are compared with version 5.96 ozone profiles retrieved by the SAGE II instrument to investigate instrumentally induced breaks and/or trends in the Uccle ozone sounding time series. This study covers the period 1985--1999, during which several changes took place in the ozone sounding procedure and processing method. These changes introduced breaks in the time series of ozone soundings, and a number of corrections have been developed that are not part of the standard processing method for ozone soundings. In this paper, we review the standard corrections applied to ozone soundings and describe additional corrections developed at Uccle. The improvement of the ozonesonde data set against SAGE II ozone measurements is shown. Jumps of up to 6% in the relative difference between SAGE II and ozonesondes are removed with corrections taking into account quality changes of the ozonesonde and the radiosonde. New corrections and refinements to the existing correction procedure reduce the drift between SAGE II and ozonesondes from -0.51% yr-1 to -0.07% yr-1 between 17 and 22 km with nonsignificant values at the 2&sgr; level at all altitudes. ¿ 2001 American Geophysical Union