In clinical practice it is of vital importance to track the health of a patient's cardiovascular system via the continuous measurement of hemodynamic parameters. Cardiac output (CO) and the related stroke volume (SV) are two such parameters of central interest as they are closely linked with oxygen delivery and the health of the heart. Many techniques exist to measure CO and SV, ranging from highly invasive to noninvasive ones. However, none of the noninvasive approaches are reliable enough in clinical settings. To overcome this limitation, we investigated the feasibility and practical applicability of noninvasively measuring SV via electrical impedance tomography (EIT), a safe and low-cost medical imaging modality. In a first step, the unclear origins of cardiosynchronous EIT signals were investigated in silico on a 4D bioimpedance model of the human thorax. Our simulations revealed that the EIT heart signal is dominated by ventricular activity, giving hope for a heart amplitude-based SV estimation. We further showed via simulations that this approach seems feasible in controlled scenarios but also suffers from some limitations. That is, EIT-based SV estimation is impaired by electrode belt displacements and by changes in lung conductivity (e.g. by respiration or liquid redistribution). We concluded that the absolute measurement of SV by EIT is challenging, but trending - that is following relative changes - of SV is more promising. In a second step, we investigated the practical applicability of this approach in three experimental studies. First, EIT was applied on 16 mechanically ventilated patients in the intensive care unit (ICU) receiving a fluid challenge to improve their hemodynamic situation. We showed that the resulting relative changes in SV could be tracked using the EIT lung amplitude, while this was not possible via the heart amplitude. The second study, performed on patients in the operating room (OR), had to be prematurely terminated due to too low variations in SV and technical challenges of EIT in the OR. Finally, the third experimental study aimed at testing an improved measurement setup that we designed after having identified potential limitations of available clinical EIT systems. This setup was tested in an experimental protocol on 10 healthy volunteers undergoing bicycle exercises. Despite the use of subject-specific 3D EIT, neither the heart nor the lung amplitudes could be used to assess SV via EIT. Changes in electrode contact and posture seem to be the main factors impairing the assessment of SV. In summary, based on in silico and in vivo investigations, we revealed various challenges related to EIT-based SV estimation. While our simulations showed that trending of SV via the EIT heart amplitude should be possible, this could not be confirmed in any of the experimental studies. However, in the ICU, where sufficiently controlled EIT measurements were possible, the EIT lung amplitude showed potential to trend changes in SV. We concluded that EIT amplitude-based SV estimation can easily be impaired by various factors such as electrode contact or small changes in posture. Therefore, this approach might be limited to controlled environments with the least possible changes in ventilation and posture. Future research should scrutinize the lung amplitude-based approach in dedicated simulations and clinical trials.