The NASA TRACE A experiment (September--October 1992) investigated effects of dry season biomass burning emissions from both South America and southern Africa on the tropical South Atlantic troposphere. Whole air canister samples were collected aboard the NASA DC-8 aircraft and analyzed for a wide range of nonmethane hydrocarbons (NMHCs) and halocarbons. Fast response in situ quantification of CH4, CO, and CO2 were also performed on the DC-8. Sampling took place over Brazilian agricultural areas and southern African savanna where there was active biomass burning. The vertical distribution of the measured gases revealed that the concentrations of most hydrocarbons, methyl halides, CH4, CO, and CO2, were enhanced in the boundary layer of these regions principally as a result of biomass fires. Brazilian and African biomass burning emission ratios were calculated for CH3Br, CH3Cl, CH3I, and NMHCs relative to CO and CO2. Although both fire regions were dominated by efficient (flaming) combustion (CO/CO2 ratios <0.1), the Brazilian samples exhibited the lower (more flaming) CO/CO2 ratio of 0.037 compared with the African savanna value of 0.062. This difference in combustion efficiency was reflected in lower ratios versus CO2 for all correlated gases. However, the gases more closely associated with smoldering combustion (e.g., C3H8 and CH3Cl) exhibited significantly higher ratios relative to CO for the Brazilian fires, while the African samples exhibited higher values versus CO for compounds associated more closely with flaming combustion (e.g., C2H2). This variation in the trace gas ratios versus CO is most likely caused by different fuel characteristics. On the basis of the emission ratios obtained, the total biomass burning emission rates for savannas and worldwide were calculated for the hydrocarbons and methyl halides. From these it was estimated that roughly 25% and 20% of global CH3Cl and CH3Br emissions, respectively, derive from biomass burning but that the contribution of biomass burning to total CH3I emissions was not significant. ¿ American Geophysical Union 1996