The spatiotemporal variability of ground-level ozone (GLO) in the rural Annapolis Valley, Nova Scotia was investigated between August 29, 2006, and September 28, 2007, using Ogawa nitrite-impregnated passive diffusion samplers (PS). A total of 353 PS measurements were made at 17 ambient and 1 indoor locations over 18 sampling periods ranging from 2 to 4 weeks. The calculated PS detection limit was 0.8 +/- 0.02 parts per billion by volume (ppbv), for a 14-day sampling period. Duplicate samplers were routinely deployed at three sites and these showed excellent agreement (R2 values of 0.88 [n = 11], 0.95 [n = 17], and 0.96 [n = 17]), giving an overall PS imprecision value of 5.4%. Comparisons between PS and automated continuous ozone analyzers at three sites also demonstrated excellent agreement with R2 values of 0.82, 0.95, and 0.95, and gradients not significantly different from unity. The minimum, maximum, and mean (+/- 1 sigma) ambient annual GLO concentrations observed were 7.7, 72.1, and 34.3 +/- 10.1 ppbv, respectively. The three highest sampling sites had significantly greater (P = 0.032) GLO concentrations than three Valley floor sites, and there was a strong correlation between concentration and elevation (R2 = 0.82). Multivariate models were used to parameterize the observed GLO concentrations in terms of prevailing meteorology at an elevated site found at Kejimkujik National Park and also at a site on the Valley floor. Validation of the multivariate models using 30 months of historical meteorological data at these sites yielded R2 values of 0.70 (elevated site) and 0.61 (Valley floor). The mean indoor ozone concentration was 5.4 +/- 3.3 ppbv and related to ambient GLO concentration by the equation: indoor = 0.34 x ambient - 5.07. This study has demonstrated the suitability of PS for long-term studies of GLO over a wide geographic area and the effect of topographical and meteorological influences on GLO in this region.