The release of carbon as CO2 from belowground processes accounts for about 70% of total ecosystem respiration. Insights about factors controlling soil CO2 efflux are constrained by the challenge of apportioning sources of CO2 between autotrophic tree roots (and mycorrhizal fungi) and heterotrophic microorganisms. In some temperate conifer forests, the reduction in soil CO2 efflux after girdling (phloem removal) has been used to separate these sources. Girdling stops the flow of carbohydrates to the belowground portion of the ecosystem, which should slow respiration by roots and mycorrhizae while heterotrophic respiration should remain constant or be enhanced by the decomposition of newly dead roots. Therefore, the reduction in CO2 efflux after girdling should be a conservative estimate of the belowground flux of C from trees. We tested this approach in two tropical Eucalyptus plantations. Tree canopies remained intact for more than 3 months after girdling, showing no reduction in light interception. The reduction in soil CO2 efflux averaged 16-24% for the 3-month period after girdling. The reduction in CO2 efflux was similar for plots with one half of the trees girdled and those with all of the trees girdled. Girdling did not reduce live fine root biomass for at least 5 months after treatment, indicating that large reserves of carbohydrates in the root systems of Eucalyptus trees maintained the roots and root respiration. Our results suggest that the girdling approach is unlikely to provide useful insights into the contribution of tree roots and heterotrophs to soil CO2 efflux in this type of forest ecosystem.