A modified residual method for deriving time-averaged tropical tropospheric ozone (TTO) from total ozone mapping spectrometer (TOMS) has been refined Kim et al., 1996> and applied to a 2-year period, 1991--1992. This period was selected because of the availability of a unique sonde record at tropical Atlantic and near-Atlantic sites: Natal, Brazil; Ascension Island; and Brazzaville. The essentials of the modified-residual method are (1) Fourier analysis to identify the latitude range for use of the method; (2) use of the wave-like pattern to estimate stratospheric ozone column and a background tropospheric ozone column, which is defined as the amount of ozone expected in the absence of chemical pollution; (3) assumption that the wave-one feature is tropospheric; and (4) the stratospheric and background tropospheric ozone amounts consistent with the constraint of the three station ozonesonde measurements for total tropospheric ozone. An excess ozone (pollution) signal is also derived from TOMS total ozone. Parameters derived from the modified residual method are evaluated with independent data to verify the quality of the TTO. Derived tropical stratospheric column ozone for 1991--1992 agrees very well with the corresponding UARS Microwave Limb Sounder and Stratospheric Aerosol and Gas Experiment records (to within 5 Dobson units (DU) for the latter). Background tropospheric ozone, which exhibits a seasonality in 1991--1992, has a 24¿4.8 DU mean value at wave maximum (~0¿ longitude) and a 15.2¿6.0 DU mean value at 180¿ longitude. The Atlantic value is remarkably similar to the climatological 26 DU previously inferred from subtropical sondes <Kim et al., 1996; Komhyr et al., 1989>. The ~5 DU standard deviation is the accuracy of the modified-residual method and is the best that can be achieved given the 15-day averaging of the sondes (1&sgr;=5.3 DU). The wave-like pattern in 1991--1992 has a semiannual periodicity with zero wave amplitude in June and December. ¿ 1998 American Geophysical Union