The 18O content of atmospheric O2 is an important tracer for past changes in the biosphere and has been used to estimate changes in the balance between terrestrial and marine productivity. Its quantitative use depends on knowledge of the isotopic fractionations associated with the various O2 production and consumption processes. Here we monitored oxygen concentration and &dgr;18O of O2 in sandy and clayey soils to evaluate in situ 18O fractionation associated with soil respiration. In the clayey soil, O2 concentrations decreased as low as 1% at 150 cm depth, and &dgr;18O values ranged from 0? to -1.6? relative to atmospheric O2. In the sandy soil the O2 concentration was 20.38--20.53%, and &dgr;18O values were -0.06¿0.015? to 0.06¿0.015? relative to atmospheric O2. Using the observed [O2] and &dgr;18O profiles and their change with time, together with a one-box analytical model and a five-box numerical model, a mean discrimination of 12¿1? was estimated for the two sites (including effects of concentration and temperature gradients). This low discrimination was consistent with that determined in closed-system soil incubation experiments (8.4--16.9?). The current understanding of the composition of air O2 attributes the magnitude of the fractionation in soil respiration to biochemical mechanisms alone (about 18? and 25--30? in cyanide-sensitive and cyanide-resistant respiration, respectively). The low discrimination we report is significantly less than in dark respiration and is explained by diffusion limitation in soil aggregates and root tissues that results in low O2 concentration in the consumption site. Soil respiration is a major component of the global oxygen uptake, and the potential contribution of low discrimination, such as observed here, to the global Dole effect should be considered in global-scale studies. ¿ 2001 American Geophysical Union