Atmospheric observations of CCl2F2 concentrations from 1976--1980 (Rasmussen et al., 1981) have been compared with atmospheric models containing stratospheric destruction alone, and with various added hypothetical tropospheric sinks ranging from 10--100 year lifetimes. Comparisons are made using (a) absolute atmospheric burdens of CCl2F2; (b) the time trend of relative concentrations; and (c) the ratio of concentrations measured in the northern and southern hemispheres. The atmospheric emissions of CCl2F2 estimated by the Chemical Manufacturers Association (CMA) fail to fit the observations with stratospheric loss alone in all three tests. The inclusion of tropospheric loss processes appreciably worsens the agreement. Three other hypothetical emissions estimates are given which satisfactorily simulate the atmospheric observations by all three tests. The emissions in these models maximize in 1978 or later at more than 500 kilotons/year, in contrast to the CMA maximum of 420 kilotons during 1974. The total emissions of CCl2F2 during 1976 through 1979 in these other models are about 35% greater than in the CMA estimates.