Eddy covariance was used to measure the net CO2 exchange (NEE) over ecosystems differing in land use (forest and agriculture) in Thuringia, Germany. Measurements were carried out at a managed, even-aged European beech stand (Fagus sylvatica, 70–150 years old), an unmanaged, uneven-aged mixed beech stand in a late stage of development (F. sylvatica, Fraxinus excelsior, Acer pseudoplantanus, and other hardwood trees, 0–250 years old), a managed young Norway spruce stand (Picea abies, 50 years old), and an agricultural field growing winter wheat in 2001, and potato in 2002. Large contrasts were found in NEE rates between the land uses of the ecosystems. The managed and unmanaged beech sites had very similar net CO2 uptake rates (∼−480 to −500 g C m−2 yr−1). Main differences in seasonal NEE patterns between the beech sites were because of a later leaf emergence and higher maximum leaf area index at the unmanaged beech site, probably as a result of the species mix at the site. In contrast, the spruce stand had a higher CO2 uptake in spring but substantially lower net CO2 uptake in summer than the beech stands. This resulted in a near neutral annual NEE (−4 g C m−2 yr−1), mainly attributable to an ecosystem respiration rate almost twice as high as that of the beech stands, despite slightly lower temperatures, because of the higher elevation. Crops in the agricultural field had high CO2 uptake rates, but growing season length was short compared with the forest ecosystems. Therefore, the agricultural land had low-to-moderate annual net CO2 uptake (−34 to −193 g C m−2), but with annual harvest taken into account it will be a source of CO2 (+97 to +386 g C m−2). The annually changing patchwork of crops will have strong consequences on the regions' seasonal and annual carbon exchange. Thus, not only land use, but also land-use history and site-specific management decisions affect the large-scale carbon balance.