We investigate the large systematic biases, especially in the stratosphere, between ozone profiles retrieved from the Global Ozone Monitoring Experiment (GOME) and ozonesonde observations at some ozonesonde stations. GOME retrievals are intercompared with both ozonesonde data at 33 stations between 75¿N and 71¿S and Stratospheric Aerosol and Gas Experiment II (SAGE II) data during 1996--1999. GOME stratospheric column ozone (SCO) over the altitude range ~15--35 km usually agrees with SAGE II SCO to within 2.5 DU (1.5%, 1 DU = 2.69 ¿ 1016 molecules cm-2) without significant spatiotemporal dependence but is systematically larger than ozonesonde SCO by 8--20 DU (5--10%) over carbon iodine (i.e., an ozonesonde technique) stations and most stations within 30¿N--30¿S. Evaluation of GOME, SAGE II, TOMS, and Dobson data here demonstrates that those biases mainly originate from ozonesonde underestimates in the stratosphere. GOME retrievals also show large positive biases of 20--70% at carbon iodine stations (except for Syowa) and most stations within 30¿N--30¿S over ~10--20 km, where ozone concentration is low, while the biases relative to SAGE II data over ~15--20 km is usually 10--20%. The discrepancies over this altitude region reflect biases in GOME retrievals as well as ozonesonde measurements. In addition, GOME/sonde biases in both SCO and profiles (especially in the lower stratosphere and upper troposphere) vary from station to station and depend on sonde technique, instrument type, sensor solution, and data processing, demonstrating the need to homogenize available ozonesonde data sets and standardize future operational procedures for reliable and consistent satellite validation.