Atmospheric Lifetime Experiment measurements of CFCl3 and CF2Cl2 several times per day at five remote surface sites from July 1978 to June 1983 are reported. For January 1, 1981, the mean latitudinally averaged mixing ratios were 177.5 parts per trillion by volume (pptv) and 300.4 pptv for CFCl3 and CF2Cl2, respectively. The atmospheric content at that time is estimated to have been 3,960 million kg of CFCl3 and 6,000 million kg of CF2Cl2. The mixing ratios exhibited an annual rate of increase of 9.0 and 15.3 pptv/yr for CFCl3 and CF2Cl2, respectively. Trend lifetime estimates for January 1, 1981, from this 5-year data set are 74+31-17 years for CFCl3 and 111+222-44 years for CF2Cl2. On the basis of a comparison of CFCl3 data on two different chromatographic columns, it is shown that the spectrum of measurement errors maximizes at low frequencies but may be relatively flat at periods longer than approximately 3 years. The uncertainties in the trend lifetime estimates are, however, dominated by release uncertainties. Inverting the analysis, assuming stratospheric photodissociation to be the only atmospheric sink of fluorocarbons, yields annual release estimates with an accuracy, based on the precision of the measurement system, of approximately 8%. Excellent agreement with the Chemical Manufacturers Association release estimates is found for CFCl3; for CF2Cl2 the estimates, although exhibiting variability from year to year, suggest that the emissions in the USSR and Eastern Europe have remained roughly constant over the years 1979--1982.