As concern over global warming intensifies, sequestration and storage of atmospheric CO2 has become an important scientific and policy issue. Confusion persists, however, over interpretation of forest carbon (C) source-sink dynamics, in part because conclusions drawn depend on temporal and spatial scales of analysis (e.g. day-week scale vs. successional-scale), type of disturbance, and methodology (e.g. massbased vs. flux-based). There is a need to resolve this confusion given that strategies for mitigating anthropogenic CO2 emissions are based on estimates of forest C fluxes during various stages of succession, over which C fluxes and stores may change. Empirical study of changes in forest C stores can help to resolve this confusion by clarifying the C sources-sink dynamics of forests in space and time. To better understand the impacts of disturbance on C source-sink dynamics, changes in C stores of an evergreen-dominated forest on the Wind River Ranger District in Southwestern Washington, U.S.A., were investigated along a 500-year chronosequence of 36 stands. Principle objectives were to evaluate 1) decomposition rates (k) of logs, stumps, and below-ground coarse roots, 2) net primary productivity (NPP) of dominant tree species' boles at the stand level, and 3) successional changes in net ecosystem productivity (NEP) for live trees and coarse woody debris (CWD), here called NEPW. In the case of decomposition, log and stump k values did not differ significantly within the two principle species studied, indicating substitution of log k values for stump k values in models of forest C budgets may be valid when stump decomposition data is lacking. Decomposition rates between species differed, with Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) decomposing more slowly (k = 0.013 yr-1) relative to western hemlock (Tsuga heterophylla (Rafi) Sarg. (k = 0.036 yr-1). This difference in k between species was observed for both above-ground stumps and logs as well as below-ground coarse roots. Given our mean k estimates and adjusting for regenerating stand age, these stands are losing C at a rate of 0.16-0.83 Mg C ha-1 yr-1 (assuming all CWD is P. menziesii) to 0.13-1.68 Mg C ha-1 yr-1 (assuming all CWD is T. heterophylla) from stumps, logs, and snags. Including coarse roots increases these losses to 0.28-1.25 Mg C ha-1 yr-1 and 0.30-2.53 Mg C ha-1 yr-1, respectively. Based on these findings, if fragmentation of these decomposing C pools is ignored, and fragmented fractions have oxidized to CO2, stands thought to be net C sinks could in reality be net C sources to the atmosphere. Net primary production in tree boles (NPPb) of regenerating stands (so called second-growth) ranged between 0.15-5.28 Mg C ha-1 yr-1. NPPb of 500-year old stands ranged between 1.3-3.9 Mg C ha-1 yr-1, similar to NPPb of boles in 20-25 year old secondgrowth. Mean radial increment widths from old-growth stands indicated that NPPb of these stands (neglecting mortality) can increase, decrease, or remain relatively constant. Based on 5-year increments for the previous fifteen years, the majority of old-growth stands sampled showed small increases in radial growth over time. Timing of the transition from negative to positive of NEPW ranged between 0 and 57 years after disturbance and depended strongly on live-tree growth rates as well as the fate of CWD and harvested wood. Estimated maximum and minimum NEPW were 3.9 Mg C ha-1 yr-1 and 14.1 Mg C ha-1 yr-1, respectively. Maximum mean C stores of 393 Mg C ha-1 were reached approximately 200 years after disturbance. At a rotation age of 80 years, regenerating stands stored approximately 50% as much C in woody biomass as a 500-year old primary forest, indicating conversion of older forests to plantations released C to the atmosphere. Given the high biomass of mature and old-growth stands relative to younger regenerating stands in the forest studied, landscape C stores in live wood would appear to be maximized in stands of older age classes.