Previous regression trend models for total column ozone have included only the quasi-biennial oscillation (QBO) winds and the El Ni¿o Southern Oscillation (ENSO) ocean surface pressure as dynamical proxies. Trends derived from these regression models generally differ (are more negative) from two-dimensional (2-D) chemical transport trends by about 2--5% decade-1 in midlatitudes during spring. The present study introduces additional dynamical proxies of total ozone in regression models in an effort to reduce errors in local ozone trends and reduce these model differences. Nimbus 7 total ozone mapping spectrometer (TOMS) version 7 total column ozone for 1979--1992 are used in conjunction with analyses from National Centers for Environmental Prediction (NCEP, formerly National Meteorological Center). Dynamical proxies investigated include winds (including diabatic winds), relative vorticity, potential vorticity, temperatures, and geopotential heights. Inclusion of additional dynamical proxies improves statistics by reducing residuals and uncertainty regions in both zonal mean and zonally asymmetric trend models. RMS reductions, relative to a trend model with only QBO, solar, and trend terms, are as large as 50% in 14-year means in the southern hemisphere. For zonal mean or zonally asymmetric global trend models with one optional surrogate, a favorable choice is prefiltered (at least deseasonalized and detrended) lower stratospheric temperatures. Relative vorticity, potential vorticity, and geopotential heights all exhibit similar relationships with total ozone, with highest correlative behavior near 200 hPa (midlatitudes year-round) and 10 hPa (high latitudes in winter-spring months). For models incorporating these latter proxies, combined 10- and 200-hPa (or similar) pressure levels are effective in reducing global residuals. ENSO, as a surrogate by itself or included with other dynamical proxies, has a comparatively small effect because of its episodic nature. Decadal variabilities in NCEP and microwave sounding unit channel 4 (MSU4) data as surrogates in trend models indicate maximal 1--3% decade-1 reductions anywhere in TOMS trends. Total ozone trends derived from the Goddard 2-D heterogeneous chemistry and transport model agree favorably with trends in TOMS ozone, generally to within 2--3% decade-1 in both hemispheres. Inclusion of possible decadal variabilities in dynamics may yield yet smaller differences.¿ 1997 American Geophysical Union