A nonhydrostatic, Boussinesq, three‐dimensional model, the ocean large‐eddy simulation model (OLEM), has been developed to study deep oceanic convection. The model uses a subgrid‐scale parameterization of turbulence developed for large‐eddy simulation models, and the advection of scalars is accomplished using a monotonic scheme. A set of experiments was performed using OLEM to provide a direct comparison with laboratory results and aircraft measurements of the atmospheric convective boundary layer. The results from these experiments are in excellent agreement with laboratory and atmospheric convective boundary layer measurements of the mean profiles of zonal and vertical velocity variance, potential temperature variance, and heat flux. The horizontal wavenumber spectra of zonal and vertical velocity are also in good agreement with laboratory measurements and Kolmogorov's theoretical inertial subrange spectrum. A set of experiments using a potential temperature‐salinity profile from the central Greenland Sea for model initialization was conducted to study the effect of the thermobaric instability and rotation on the structure and evolution of deep oceanic convection. The artificial removal of the thermobaric instability suppresses penetrative convection, which is responsible for rapid changes in water properties at depths much greater than occurs for convective, mixed‐layer deepening. The vertical velocity and diameter, −0.08 m s−1 and 300 m, respectively, of the penetrative plumes are in good agreement with observations from the Greenland Sea. A period of strong penetrative convection is followed by a gradual transition to convective, mixed‐layer deepening. During penetrative convection, the values of heat flux are about 2 times greater than convective, mixed‐layer deepening. In the absence of rotation, the evolution of penetrative convection occurs more rapidly, and vertical motions are more vigorous. The presence of the horizontal component of rotation forces asymmetries in the circulation around a penetrative plume. These experiments clearly demonstrate the importance of thermobaric instability and rotation on deep convection. To properly model large‐scale flows in regions of penetrative convection, it is necessary to include these effects in the vertical mixing parameterization.