Analysis of the spatial correlation of ozone mixing ratio in the vertical provides information useful for several purposes: (a) it aids description of the degree of regionality of the ozone transport-transformation processes, (b) the information provided in the form of a priori covariance matrices for remote retrieval algorithms can simplify and sharpen accuracy of the resulting estimates, and most importantly, (c) it allows a first evaluation of the improvement that remote retrievals can give over boundary-layer climatology. Vertical profiles of mean, variance, and vertical autocovariance, and vertical autocorrelation of ozone mixing ratios were estimated and given parameterizations. The WOUDC ozonesonde network database was used. During the years 2004–2006, these were considerably augmented by sondes taken by NASA, NOAA, and Canadian agencies during recent summertime intensive periods in North America. There are large differences across the North American continent in the patterns and magnitudes of correlation, especially in the lowest 2–3 km of the troposphere. This is especially significant for the near-surface layers (100's of meters deep) which determine actual surface O3 smog exposure and phytotoxicity, since satellite retrievals typically characterize at best a thick layer extending 3 km or more from the surface. The relative variation of O3 decreases in the vertical, particularly for the somewhat polluted launch stations, and this affects inference of surface O3 significantly. We outline a simple synthesis of mixed-layer and ozone-chemistry behavior to aid discussion of this and similar phenomena. Regional differences suggest broad if qualitative explanations in terms of larger-scale (interstate-transport) and local-scale phenomena (lake and sea breezes, degree/frequency of subsidence), inviting future study. The character of near-surface-to-full-layer covariance suggests that remote retrieval can describe surface ozone surprisingly well using 0–3 km partial-column ozone… for many situations. This indicates that there is substantial utility for new remote-retrieval methods that exploit ozone absorption in multiple wavelength regions, e.g., UV + Vis, UV + IR, or UV + Vis + IR. In summary, we find considerable value in interpreting retrievable O3 columns to estimate O3 quantities that are closely relevant to air pollution mitigation.