Fluxes of N2O and CH4 were measured monthly from 1990 through 1992 along a topogradient in a spruce forest (Picea abies L.), beech forest (Fagus silvatica L.), riparian grassland, coastal grassland, abandoned farmland, upland arable soil, and drained arable soil in Denmark. Spatial coefficints of variation (CV) in N2O emissions ranged between 106 and 617%. A consistent spatial pattern occurred across a temporary flood water boundary in the coastal grassland. Although inorganic N levels were low at this site, N2O emission rates compared with those at the N‐fertilized sites. Maximum N2O emission generally occured in the spring and autumn. Spatial CVs in CH4 fluxes ranged between 166 and 1787%, and consistent spatial patterns were found at the riparian and coastal grassland sites where CH4 was produced in temporarily flooded areas at rates of 7877 and 112 mg C m−2 yr−1, respectively. Uptake, as well as emission of CH4, reached maximum rates when soils dried up, presumably because CH4 diffusion became unconstrained. Annual site means of N2O emission were inversely correlated with CH4 uptake, and were 77 and 91 mg N and C m−2 in the spruce forest, 80 and 37 mg N and C m−2 in the beech forest, 66 and 69 mg N and C m−2 in the riparian, 124 and 7 mg N and C m−2 in the coastal grassland, 26 and 80 mg N and C m−2 in the abandoned farmland, 361 and 22 mg N and C m−2 in the upland arable, and 467 and 11 mg N and C m−2 in the drained arable. In this study, spatial variability in trace‐gas fluxes was high and poorly explained at all sites. The results suggest large‐scale integrative measurement techniques are necessary to obtain accurate flux estimates of N2O and CH4 from these ecosystems.