This paper presents the first detailed characterization of seasonal and interannual variability in tropical tropospheric column ozone (TCO) to delineate the relative importance of biomass burning and large-scale transport. TCO time series are derived from 20 years (1979--1998) of total ozone mapping spectrometer (TOMS) data using the convective cloud differential (CCD) method. Our study identifies three regions in the tropics with distinctly different characteristics related to seasonal and interannual variability. These three regions are the eastern Pacific, Atlantic, and western Pacific. TCO in the Atlantic region peaks at about the same time (September-October) both north and south of the equator, while the annual-cycle amplitude in TCO varies from about 3 to 6 Dobson units (DU) from north to south of the equator. In comparison, annual cycles in both the eastern and western Pacific are generally weak, with the largest TCO amount occurring around March-April in the northern hemisphere and September-November in the southern hemisphere. Interannual variabilities in these three regions are also very different. The Atlantic region indicates a predominant quasi-biennial oscillation (QBO) in TCO which is out of phase with the QBO in stratospheric ozone. This behavior is consistent with a UV modulation of upper tropospheric photochemistry on a QBO timescale caused by the QBO in stratospheric ozone. However, photochemical models predict significantly smaller changes, and dynamical effects may be a critical factor. Interannual variability in TCO in the Atlantic also appears to have some influence from biomass burning, but our study indicates that it is of lesser significance. Interannual variability in the eastern and western Pacific is dominated by El Ni¿o events. During an El Ni¿o there is anomalously low TCO in the eastern Pacific and high values in the western Pacific. These signatures indicate combined effects of convectively driven transport and tropospheric ozone generated by intense biomass burning in the Indonesian region. Finally, a simplified tropospheric ozone residual (STOR) method which utilizes the small variability of TCO near the dateline is proposed in this study to derive high-resolution maps and extended time series of TCO.