The convection scheme used at the UK Meteorological Office in large-scale numerical models is described. The scheme uses a “bulk” cloud model to represent an ensemble of convective clouds and aims to represent shallow, deep and midlevel convection. A simple closure is employed, the initial convective mass flux being related to the stability of the initial convecting layer. The ability of the scheme to represent convective processes in a variety of situations is evaluated using GATE, BOMEX, and ATEX data. In each case realistic heating rates are simulated and although the closure of the scheme does not demand a balance between convective and large-scale forcings as in many other types of convection scheme (for example the Arakawa–Schubert scheme), a quasi-equilibrium is established while retaining realistic atmospheric structure. The performance of the scheme in an 11-layer atmospheric general circulation model used in climate research at the UK Meteorological Office is also evaluated by comparing... Abstract The convection scheme used at the UK Meteorological Office in large-scale numerical models is described. The scheme uses a “bulk” cloud model to represent an ensemble of convective clouds and aims to represent shallow, deep and midlevel convection. A simple closure is employed, the initial convective mass flux being related to the stability of the initial convecting layer. The ability of the scheme to represent convective processes in a variety of situations is evaluated using GATE, BOMEX, and ATEX data. In each case realistic heating rates are simulated and although the closure of the scheme does not demand a balance between convective and large-scale forcings as in many other types of convection scheme (for example the Arakawa–Schubert scheme), a quasi-equilibrium is established while retaining realistic atmospheric structure. The performance of the scheme in an 11-layer atmospheric general circulation model used in climate research at the UK Meteorological Office is also evaluated by comparing...