We summarize recent numerical results on the control of the star formation efficiency (SFE), addressing the effects of turbulence and the magnetic field strength. In closed-box numerical simulations, the effect of the turbulent Mach number depends on whether the turbulence is driven or decaying: In driven regimes, increasing with all other parameters fixed decreases the SFE, while in decaying regimes the converse is true. The efficiencies in non-magnetic cases for realistic Mach numbers 10 are somewhat too high compared to observed values. Including the magnetic field can bring the SFE down to levels consistent with observations, but the intensity of the magnetic field necessary to accomplish this depends again on whether the turbulence is driven or decaying. In this kind of simulations, a lifetime of the molecular cloud (MC) needs to be assumed, being typically a few free-fall times. Further progress requires determining the true nature of the turbulence driving and the lifetimes of the clouds. Simulations of MC formation by large-scale compressions in the warm neutral medium (WNM) show that the generation of the clouds' initial turbulence is built into the accumulation process that forms them, and that the turbulence is driven for as long as accumulation process lasts, producing realistic velocity dispersions and also thermal pressures in excess of the mean WNM value. In simulations including self-gravity, but neglecting the magnetic field and stellar energy feedback, the clouds never reach an equilibrium state, but rather evolve secularly, increasing their mass and gravitational energy until they engage in generalized gravitational collapse. However, local collapse events begin midways through this process, and produce enough stellar objects to disperse the cloud or at least halt its collapse before the latter is completed. Simulations of this kind including the missing physical ingredients should contribute to a final resolution of the MC lifetime and the origin of the low SFE problems.