A three-dimensional cloud-resolving model is used to simulate the transport of lower-tropospheric passive tracers into the lowermost stratosphere via midlatitude convection. In previous studies of troposphere-to-stratosphere convective transport the extent of irreversible transport is unclear because the tropopause location is difficult to determine in the highly perturbed environment directly above an active storm. To determine the irreversibility of cross-tropopause transport in this study, 10-hour simulations are carried out to cover the growth and decay cycles of the storm. After the decay of convection, isentropes relax to quasi-flat surfaces, and the position of the tropopause becomes much easier to establish. Air parcels containing boundary layer tracers were able to penetrate the stable stratosphere because diabatic processes increased the parcel's potential temperature sufficiently to make the parcel neutrally buoyant at stratospheric altitudes. The boundary layer tracer was carried upward in the core of the updraft whereas tracers originating from higher levels were lifted on the flanks of the updraft and therefore underwent less transport into the stratosphere. Three different cases were simulated: a prototypical supercell, a prototypical multicell, and a supercell observed during the Severe Thunderstorm Electrification and Precipitation Study (STEPS) field campaign. In the prototypical supercell simulation, at 1 km above the tropopause the maximum concentration of boundary layer tracer is diluted to 26% of its original concentration; the maximum concentration of the tracer originating in the layer between 1 and 4 km is diluted to 23% of its original concentration. Simulation of the STEPS storm showed similar irreversible transport in a less idealized case. Both supercell storms produced more transport than the prototypical multicell storm.