Abstract The main objective of the present study was to determine the contributions of autotrophic and heterotrophic components to the total soil CO2 efflux over three years with high-frequency data acquisition by means of automated measurements. Soil CO2 efflux was measured continuously by using an automated open system of 10 soil respiration chambers in a sandy grassland in Hungary. Mesh-collar technique was applied to separate the components of the total respiration. Data were collected (1) in root-exclusion (Exr), (2) in root- and mycorrhiza exclusion (Exrm) and (3) in control plots (Exc, roots and mycorrhiza included). We fitted three different models to describe the dependence of total soil CO2 efflux measured on the Exc (Rs), CO2 efflux measured on Exr (RTR) and CO2 efflux measured on Exrm (RTRM) on abiotic and biotic drivers. The best fitted model (based on AIC) was later on used in a simulation process. The contribution by rhizospheric respiration (simulated, Rrhizo*) was 36 ± 21%, the contribution by mycorrhizal respiration (simulated, Rmyc*) to the total soil respiration was 9 ± 9% while the contribution by heterotrophic respiration (simulated, Rhet*) was 55 ± 21% on average. Measured mycorrhizal respiration (RM) responded to GPP with a time lag of 0–2 days in active period. Drought affected the autotrophic component of soil respiration the most intensively: raise of soil water content resulted in increase of RS by 175% while RTR and RTRM increased by 127% and 93%, respectively. The results highlight the fact that it would be useful to establish and apply separate models for each component.