A theory of ecosystem succession relates the continuum of fire frequency and intensities to mean annual carbon burning in major ecosystems of the world. Low fire frequency and release of C are contrasted with combinations of (1) low frequency, high release, (2) high frequency and release, and (3) high frequency, sometimes lowering mean C release and/or ecosystem productivity. Experience, literature, income-loss equations, matrix multiplications of probabilities, and Honolulu SCOPE workshop discussions suggest the following consequences: (1) Infrequent but drastic burns restart many stand developments and successions. (2) Frequent, intense burning regimes fundamentally modify ecosystem processes and -2 -1 composition (e.g., releasing >100 C m yr in some seasonally dry ecosystems). (3) Frequent burns, lowering available C per fire, may have as high, or lower, average burn per year (frequency x mean C burn per fire). Charts, tables, and a map of averaged annual C release, and multiplications by tentative area estimates now suggest that nonfossil C release by fire is at least slightly below the 5 x 10"^ g C yr ^ recently released by burning of fossil fuels, as global CO2 fluxes.