Partitioning soil carbon dioxide (CO2) efflux (RS) into autotrophic (RA; including plant roots and closely associated organisms) and heterotrophic (RH) components has received considerable attention, as differential responses of these components to environmental change have profound implications for the soil and ecosystem C balance. The increasing number of partitioning studies allows a more detailed analysis of experimental constraints than was previously possible. We present results of an exhaustive literature search of partitioning studies and analyse global trends in flux partitioning between biomes and ecosystem types by means of a metaanalysis. Across all data, an overall decline in the RH/RS ratio for increasing annual RS fluxes emerged. For forest ecosystems, boreal coniferous sites showed significantly higher (P<0.05) RH/RS ratios than temperate sites, while both temperate or tropical deciduous forests did not differ in ratios from any of the other forest types. While chronosequence studies report consistent declines in the RH/RS ratio with age, no difference could be detected for different age groups in the global data set. Different methodologies showed generally good agreement if the range of RS under which they had been measured was considered, with the exception of studies estimating RH by means of root mass regressions against RS, which resulted in consistently lower RH/RS estimates out of all methods included. Additionally, the time step over which fluxes were partitioned did not affect RH/RS ratios consistently. To put results into context, we review the most common techniques and point out the likely sources of errors associated with them. In order to improve soil CO2 efflux partitioning in future experiments, we include methodological recommendations, and also highlight the potential interactions between soil components that may be overlooked as a consequence of the partitioning process itself.