Low-level direct electrical current has shown promise as a potential therapeutic modality (direct current therapy; DCT) in the treatment of malignant disease, including metastases, but to date much experimental work has been empirical and has added little to our knowledge of the mechanisms involved. As a prerequisite to a clinical trial for metastases in the liver, we have employed an in vivo liver model to examine the quantitative and qualitative relationships between electrode polarity, charge and tissue necrosis. Two distinct regions of necrosis were induced, distinguishable histologically and by magnetic resonance imaging: (i) a cylindrical region of primary necrosis centred on the electrode, its volume directly proportional to the charge passed, but greater at the anode than cathode; and (ii) a wedge-shaped infarct, apex at the electrode and base extending to the liver edge. The extent of this infarct was again greater at the anode than the cathode, but showed a sigmoidal relationship with charge. Results indicate pH changes at the electrodes as likely mediators of tissue injury, but show also that significant distant ischaemic injury can occur as a consequence of primary damage. These findings should be considered when selecting tumours for possible direct current therapy and when determining the sites of electrode placement.