Aim Providing a quantitative overview of ecosystem functioning in a three-dimensional space defined by ecosystem stocks, fluxes and rates, across major ecosystem types and climatic zones.Location Global.Time period 1966-2019.Major taxa studied Ecosystem-level measurements (all organism types).Methods We conducted a global quantitative synthesis of a wide range of ecosystem variables related to carbon stocks and fluxes. We gathered a total of 4,479 values from 1,223 individual sites (unique geographical coordinates) reported in the literature (604 studies), covering ecosystem variables including biomass and detritus stocks, gross primary production, ecosystem respiration, detritus decomposition and carbon uptake rates, across eight major aquatic and terrestrial ecosystem types and five broad climatic zones (arctic, boreal, temperate, arid and tropical). We analysed the relationships among variables emerging from the comparisons of stocks, fluxes and rates across ecosystem types and climates.Results Within our three-dimensional functioning space, average ecosystems align along a gradient from fast rates-low fluxes and stocks (freshwater and pelagic marine ecosystems) to low rates-high fluxes and stocks (forests), a gradient that we hypothesize results mainly from variation in primary producer characteristics. Moreover, fluxes and rates decrease from warm to colder climates, consistent with the metabolic theory of ecology. However, the strength of climatic effects differs among variables and ecosystem types, resulting, for instance, in opposing effects on net ecosystem production between terrestrial and freshwater ecosystems (positive versus negative effects).Main conclusions This large-scale synthesis provides a first quantified cross-ecosystem and cross-climate comparison of multivariate ecosystem functioning. This gives a basis for a mechanistic understanding of the interdependency of different aspects of ecosystem functioning and their sensitivity to global change. To anticipate responses to change at the ecosystem level, further work should investigate potential feedbacks between ecosystem variables at finer scales, which involves site-level quantifications of multivariate functioning and theoretical developments.