Soil CO2 production is a key process in ecosystem C exchange, and global change predictions require understanding of how ecosystem disturbance affects this process. We monitored CO2 levels in soil gas and as bicarbonate in drainage from an experimental red pine ecosystem, for 1 year before and 3 years after its aboveground biomass was removed. Lack of physical disturbance, strict prevention of plant regrowth, and a comparison ecosystem without rooted plants facilitated isolation of the microclimatic and biochemical effects of instantaneous canopy removal and cessation of photosynthesis. Preharvest gas-phase CO2 levels fluctuated with growing-season soil temperature but reached their greatest levels (up to 10,000 ppmV) during late winter beneath snow and ice cover. This pattern, and the annual CO2 efflux of ~500 g C m-2 yr-1, continued for 2 years following harvest; the efflux declined by half in the third year. The surprising continuity of preharvest and postharvest rates of soil CO2 production reflects the replacement of root respiration with microbial respiration of root and litter substrates of declining lability, but boosted by soil temperature increases. Mass balance is consistent with a bulk root+litter exponential decay time (-1/k) of 4--6 years, such that most of the subsurface biomass accumulated over 15 years of tree growth would be lost in a decade after the harvest. The preharvest bicarbonate C efflux, which was less than 0.1% of the gas-phase efflux, trebled after the harvest owing to elimination of evapotranspiration and consequent increases in drainage while soil CO2 levels remained high. A large fraction of this "hydrospheric" sink for atmospheric CO2 is attributed to weathering under high soil CO2 levels before spring snowmelt and soil-water flushing. These observations suggest that disturbance may enhance long-term chemical-weathering CO2 sinks.