Carbon dioxide partial pressure (pCO₂) in surface seawater was continuously recorded every three hours from 18 July 2006 through 31 October 2007 using a moored autonomous pCO₂ (MAPCO₂) system deployed on the Gray’s Reef buoy off the coast of Georgia, USA. Surface water pCO₂ (average 373 ± 52 μatm) showed a clear seasonal pattern, undersaturated with respect to the atmosphere in cold months and generally oversaturated in warm months. High temporal resolution observations revealed important events not captured in previous ship-based observations, such as sporadically occurring biological CO₂ uptake during April–June 2007. In addition to a qualitative analysis of the primary drivers of pCO₂ variability based on property regressions, we quantified contributions of temperature, air–sea exchange, mixing, and biological processes to monthly pCO₂ variations using a 1-D mass budget model. Although temperature played a dominant role in the annual cycle of pCO₂, river inputs especially in the wet season, biological respiration in peak summer, and biological production during April–June 2007 also substantially influenced seawater pCO₂. Furthermore, sea surface pCO₂ was higher in September–October 2007 than in September–October 2006, associated with increased river inputs in fall 2007. On an annual basis this site was a moderate atmospheric CO₂ sink, and was autotrophic as revealed by monthly mean net community production (NCP) in the mixed layer. If the sporadic short productive events during April–May 2007 were missed by the sampling schedule, one would conclude erroneously that the site is heterotrophic. While previous ship-based pCO₂ data collected around this buoy site agreed with the buoy CO₂ data on seasonal scales, high resolution buoy observations revealed that the cruise-based surveys undersampled temporal variability in coastal waters, which could greatly bias the estimates of air–sea CO₂ fluxes or annual NCP, and even produce contradictory results.