Current dosimetry protocols recommend the use of plane-parallel ionization chambers for the dosimetry of clinical electron beams. The necessary perturbation corrections p wall and p cav are assumed to be unity, independent of the depth of measurement and the energy of the primary electrons. To verify these assumptions detailed Monte Carlo studies of a Roos chamber in clinical electron beams with energies in the range of 6–21 MeV are performed at different depths in water and analyzed in terms of Spencer–Attix cavity theory. Separate simulations for the perturbation corrections p wall and p cav indicate quite different properties of both correction factors with depth. Dose as well as fluence calculations show a nearly depth-independent wall correction factor for a shift of the Roos chamber Δz=− 0.017 cm toward the focus. This value is in good agreement with the positioning recommendation given in all dosimetry protocols. Regarding the fluence perturbation p cav the simulation of the electron fluence inside the air cavity in comparison to water unambiguously reveals an in-scattering of low energy electrons, despite the fact, that the cavity is' well guarded'. For depths beyond the reference depth z ref this effect is superimposed by an increased loss of primary electrons from the beam resulting in p cav> 1. This effect is largest for low electron energies but present for all electron energies involved in this study. Based on the different depth dependences of p wall and p cav it is possible to choose a chamber shift Δz in a way to minimize the depth dependence of the overall perturbation factor p. For the Roos chamber this shift is Δz=− 0.04 cm independent of electron energy.