It is believed that in the quasigeostrophic framework, the height of the midlatitude tropopause is tantamount to the penetration depth of the potential vorticity flux due to baroclinic eddies (Held 1982). The penetration depth has hitherto been associated largely with (quasi) linear theories and with baroclinic adjustment hypothesis (e.g., Green 1970, Stone 1972, Lindzen and Farrell 1980). We re-examine the parametrical dependence of the penetration depth in the baroclinic turbulence regime, where the relevance of linear theory and baroclinic adjustment is unfounded. A theoretical prediction based on a simple diffusive flux closure will be compared with direct numerical simulations of quasigeostrophic baroclinic turbulence of stratified flow.