In New England, 70 to 90% of the lands once cleared for agriculture are now mature forests. These aggrading temperate forests sequester significant soil organic carbon (SOC) during regeneration as they return to pre-disturbance equilibriums. Numerous studies have estimated rates of SOC sequestration by calculating the difference between the aggrading forest and an assumed SOC pool for the previous agricultural land and dividing the value by the age of the forest. Few studies, however, have used process level measures of carbon losses and additions to support these sequestration rates. In this study, we focused on the flux of SOC in southern New England forests recovering from agricultural abandonment and the processes governing net sequestration at a small watershed scale. Processes governing SOC sequestration were examined in three forested watersheds, ranging in size from 32 to 142 ha. Additions of SOC were attributed to leaf litter, deadfall, and roots. In each watershed: leaf litter was collected in 1430 cm² trays distributed at 14 locations; deadfall additions were collected from 10 plots each having an area of 2500 cm²; and fine roots were collected from 18 buried simulated peds. The peds were constructed of root-free soil encased in nylon bags to a size of 60 cm³. Each simulated ped was buried before the growing season to a depth of 5 – 20 cm and collected after the growing season ended. Coarse root additions were estimated to be equivalent to the amount of fine roots we measured in the simulated peds. Losses to the SOC pool were measured as soil CO² respiration and dissolved organic carbon (DOC) transport. Soil CO² respiration was measured at 12 locations in each watershed on a monthly-basis using a infrared gas analyzer. Stream samples were taken at the point where the streams exited the watershed and analyzed for DOC content. Stream discharge rates were determined using fluorometric dye tracing techniques for various stream heights to develop a stage-discharge relationship. Stream DOC concentrations were multiplied by corresponding stream discharges to determine stream DOC loss throughout the year. Leaf litter, deadfall, and root biomass contributed on average 7.24 Mg C ha^-1 within one of the watersheds. Losses of SOC totaled 6.95 Mg C ha^-1, with almost all of the losses in the form of soil CO² respiration (6.91 Mg C ha^-1). Most of these losses (81%) occurred from May until October. Mass balance calculations indicate a net SOC sequestration of 0.29 Mg C ha^-1 yr^-1 over the data collection period. As a comparison to the process level sequestration rate, we examined a chronosequence of nine forests recovering from agricultural abandonment ranging from 27 to 86 years. To account for previous land-use practices, soils of the same type were sampled from paired sites containing previously cultivated forests and adjacent agriculture fields still under cultivation. Sites were separated into two land-use change categories: agriculture to deciduous forest and agriculture to coniferous forest and differences in SOC between the forest and field were calculated. Soil carbon sequestration rates for all sites ranged from 0.19 to 2.08 Mg C ha^-1 yr^-1. The mean sequestration rate for all sites was 0.83 Mg C ha^-1 yr^-1. Mean carbon sequestration rates for coniferous sites were significantly higher (p = 0.05) than mean rates for deciduous sites (1.09 and 0.52 Mg C ha^-1 yr^-1, respectively). The highest rate observed for deciduous sites (0.71 Mg C ha^-1 yr^-1) was the lowest rate recorded for coniferous sites. Although the process based, watershed level SOC sequestration rates are comparable to rates calculated using the chronosequence approach (0.29 vs 0.83 Mg C ha^-1 yr^-1), the lower average value of the former suggests a possible imbalance in SOC sequestration rates. One explanation for the difference may be in our inability to directly measure coarse root additions, which we estimated as equivalent to fine root additions.