Methyl bromide (MeBr) is used increasingly as a biocidal fumigant, primarily in agricultural soils prior to planting of crops. This usage carries potential for stratospheric ozone reduction due to Br atom catalysis, depending on how much MeBr escapes from fumigated soils to the atmosphere and on details of atmospheric chemical reactions. We present direct field measurements of MeBr escape; 87% of the applied MeBr was emitted within 7 days after a commercial fumigation. Covering the field with plastic sheets retarded MeBr escape somewhat but first-day losses were still 40%; thicker sections of sheets were relatively more effective than thin sections. We also measured gaseous MeBr concentrations versus depth in the soil column; these profiles display diffusion-like evolution. In soil, MeBr is partitioned among gas, liquid, and adsorbed solid phases. Calculated soil inventories agreed only roughly with applied amounts, probably due to nonequilibrium partitioning (during the first 30 min) and to uncertainties in partitioning coefficients. Fumigated fields may release less MeBr if they are covered by more gas-tight plastic films, if injection techniques are improved and injection is deeper, and if soil moistures, organic amounts, and densities are greater than in the soil studied here.