Abstract: The unique and distinctive 17O-excess (D17O) of ozone (O3) provides a conservative tracer for oxidative processes in both modern and paleo-atmospheres and has acted as the primary driver of theoretical and experimental research into non-mass-dependent fractionation (NMDF) for over three decades. However, due to the inherent complexity of extracting O3 from ambient air, the existing observational dataset for tropospheric O3 isotopic composition remains quite small. Recent analytical developments have provided a robust and reliable means for determining D17O(O3)trans., the transferrable D17O signature of ozone in the troposphere (Vicars et al., 2012). We have employed this new methodology in a systematic investigation of the spatial and seasonal features of D17O(O3)trans. in two separate field campaigns: a weekly sampling effort at our laboratory in Grenoble, France (45°N) throughout 2012 (n = 47) and a four-week campaign onboard the Research Vessel (R/V) Polarstern along a latitudinal transect from 50°S to 50°N in the Atlantic Ocean (n = 30). The bulk 17O-excess of ozone, denoted D17O(O3)bulk, exhibited mean (±1r) values of 26.2 ± 1.3& (D17O(O3)trans. = 39.3 ± 2.0&) and 25.9 ± 1.1& (D17O(O3)trans. = 38.8 ± 1.6&) for the Grenoble and R/V Polarstern collections, respectively. This range of values is in excellent quantitative agreement with the two previous studies of ozone triple-isotope composition, which have yielded mean (±1r) D17O(O3)bulk values of 25.4 ± 9.0& (n = 89). However, the magnitude of variability detected in the present study is much smaller than that formerly reported. In fact, the standard deviation of D17O(O3)bulk in each new dataset is lower than the uncertainty previously estimated for the filter technique (±1.7&), indicating a low level of natural spatial and temporal variation in the 17O-excess of surface ozone. For instance, no clear temporal pattern in D17O(O3) is evident in the annual record from Grenoble despite dramatic seasonal variations in ozone and atmospheric reactive nitrogen (NOx = NO + NO2) concentrations. However, a small but statistically significant difference is distinguishable in the R/V Polarstern record when comparing samples collected in the Southern and Northern Hemispheres, which possessed average D17O(O3)bulk values of 25.2 ± 1.0& and 26.5 ± 0.7&, respectively. The implications of these results are discussed in the context of the tropospheric ozone budget and the use of oxygen isotope ratios of secondary atmospheric species to derive information regarding oxidation pathways from modern and paleo-atmospheres.