We measured soil respiration and soil carbon stocks, as well as micrometeorological variables in a chronosequence of deciduous forests in Wisconsin and Michigan. The chronosequence consisted of (1) four recently disturbed stands, including a clearcut and repeatedly burned stand (burn), a blowdown and partial salvage stand (blowdown), a clearcut with sparse residual overstory (residual), and a regenerated stand from a complete clearcut (regenerated); (2) four young aspen (Populus tremuloides) stands in average age of 10 years; (3) four intermediate aspen stands in average age of 26 years; (4) four mature northern hardwood stands in average age of 73 years; and (5) an old-growth stand approximately 350-years old. We fitted site-based models and used continuous measurements of soil temperature to estimate cumulative soil respiration for the growing season of 2005 (days 133–295). Cumulative soil respiration in the growing season was estimated to be 513, 680, 747, 747, 794, 802, 690, and 571 g C m−2 in the burn, blowdown, residual, regenerated, young, intermediate, mature, and old-growth stands, respectively. The measured apparent temperature sensitivity of soil respiration was the highest in the regenerated stand, and declined from the young stands to the old-growth. Both, cumulative soil respiration and basal soil respiration at 10 °C, increased during stand establishment, peaked at intermediate age, and then decreased with age. Total soil carbon at 0–60 cm initially decreased after harvest, and increased after stands established. The old-growth stand accumulated carbon in deep layers of soils, but not in the surface soils. Our study suggests a complexity of long-term soil carbon dynamics, both in vertical depth and temporal scale.