This study describes a novel approach to separate three soil carbon (C) sources by one tracer method (here 13C natural abundance). The approach is based on the combination of C3 and C4 sources in different treatments, identical decomposition of C3 and C4 substances in soil, and subsequent calculation of their contribution to the total CO2 efflux. We used the temporal dynamics of the CO2 efflux from a C3 grassland soil amended with added C3 or C4 slurry and/or C3 or C4 sugar to estimate contributions of three separate C sources: native soil organic matter (SOM), slurry and sugar, to CO2 efflux. Soil with slurry and/or sugar was incubated under controlled conditions, and concentration and δ13C values of evolved CO2 were measured over a 2-week period. The main assumption needed for separation of three C sources in CO2 efflux, i.e. identical decomposition of applied C3 and C4 sugars in soil, was investigated and proven. The relative contribution to the CO2 efflux increased, but its duration decreased with an increased microbial availability of the C source, i.e. sugar > slurry > SOM. The microorganisms used the C sources according to their availability. The contribution of sugar to the CO2 efflux was finished after 2–4 days. Separation of three CO2 sources and comparison of CO2 from different treatments tracing the changes of SOM and slurry decomposition induced by addition of sugar were investigated. During the sugar decomposition (the first 2–4 days), the SOM decomposition strongly decreased. At the same time the contribution of slurry-C to CO2 increased. The shortcomings and limitations as well as possible future applications of the suggested method including FACE (Free Air Carbon dioxide Enrichments) and continuous labelling experiments are discussed. Copyright © 2005 John Wiley & Sons, Ltd.