Careful choice of of method, problem, and zoning has allowed us to do three-dimensional (3D) simulations of thermally relaxed, nearly adiabatic convection (with nuclear burning). The simulations are run long enough so that a robust statistical state is found. We find that 2D simulations are biased relative to 3D simulations: 2D shows larger velocities and less mixing than their 3D counterparts. Detailed theoretical analysis of these numerical experiments allows us to begin to build a simple theoretical model of turbulent convection in stars, which may be used in 1D calculations of stellar evolution. Implications for stellar evolution, will be discussed. Oxygen shell burning simulations in 3D, and multishell burning of C, Ne, O, and Si in 2D will be presented, as will aspherical distortion in supernovae progenitors (Meakin and Arnett, 2006a). Contact will be made with convective driving of waves, convective zone growth by entrainment, the velocity scale and the geometric parameters in mixing length theory, and the solar Ne abundance problem. Explicit comparisons of compressible and anelastic methods at modest Mach numbers (M ≈ 0.01 to 0.1), as well as solutions of the nonradial wave equations, are presented here. Additional detail is presented in the poster by Meakin.