Changes in land use between 1850 and 1980 are estimated to have increased the global areas in croplands, pastures, and shifting cultivation by 891, 1308, and 30 x 10^6 ha, respectively, reducing the area of forests by about 600 x 10^6 ha, releasing about 100 PgC to the atmosphere, and transferring about 23 PgC from live vegetation to dead plant material and wood products. Another 1069 X 10^6 ha are estimated to have been logged during this period, and the net release of carbon from the combined processes of logging and regrowth contributed 23 PgC to the 100-PgC release. Annual rates of land-use change and associated emissions of carbon have decreased over the last several decades in temperate and boreal zones and have increased in the tropics. The average release of carbon from global changes in land use over the decade of the 1980s is estimated to have been 1.6 +/- 0.7 PgC y-1 almost entirely from the tropics. This estimate of carbon flux is higher than estimates reported in recent summaries because it is limited here to studies concerned only with changes in land use. Other recent analyses, based on data from forest inventories, have reported net accumulations of carbon as high as 1.1 PgC y-1 in temperate and boreal zones. Because the accumulation of carbon in forests may result from natural processes unrelated to land-use change, estimates based on these inventories should be distinguished from estimates based on changes in land use. Both approaches identify terrestrial sinks of carbon. The argument is made here, however, that differences between the two approaches may help identify the location and magnitude of heretofore 'missing' sinks. Before different estimates can be used in this way, analyses must consider similar geographical regions and dates, and they must account for the accumulation and loss of carbon in forest products in a consistent fashion.
