In this deep hyperthermia study, the robustness of SAR (specific absorption rate) patterns to patient-position variations is assessed, as well as the possibilities to correct for improper positioning and the benefits of non-standard positions. With a finite element model, the SAR distributions were predicted for ten patients at 33 positions. Position sensitivity is assessed for both SAR-focus steering, i.e. settings based on a calculated focus in a cylindrical patient representation, and HTP (hyperthermia treatment planning)-guided steering, i.e. model-based optimization of the SAR distribution. Position inaccuracies of less than 1 cm do not significantly affect SAR patterns. For SAR-focus steering, the SAR maximum is not always at the desired focus location, especially in the Y (anterior/posterior)- and Z (axial)-directions. For a maximum shift of 5 cm in all directions, both SAR-focus steering and HTP-guided steering are suitable to correct for improper positioning up to the level that none of the investigated positions appears preferable. Current positioning precision is sufficient in the X (right-left)-direction, but precision measurements are needed to reach the desired accuracy in the Y-direction. In the Z-direction, a cranial shift of the applicator is predicted to be beneficial. If the position is known accurately, correction of the treatment setting is possible without loss of heating efficiency. Additionally, no preferable positions exist.