Reactive oxygen species (ROS) such as superoxide anion radical, singlet oxygen and  hydrogen peroxide appear to be important for the promotion, initiation, progression and maintenance of cancer cells.  ROS signaling plays important roles in normal biological functions, such as promoting cell proliferation.  Increases in ROS production can also lead to increased genome instability through damage to DNA and DNA associated proteins such as chromatin.  Cancer cells in some instances may down-regulate antioxidant enzymes (which are important in reducing ROS levels) for the purpose of increasing ROS levels for driving cellular proliferation.  Other cancers may up-regulate antioxidant enzymes to help protect against damage from ROS.  These adaptive responses are probably dependent upon ROS levels which probably vary between cancer cell types.  Since increased levels of ROS in cancer cells is seen as a bad thing, many therapies (such as enhancing ROS scavenging enzymes)  in development are focused on reducing ROS levels.  This strategy is hoped to reduced genomic instability which could lead to more aggressive cancers and reduce proliferation in these cells.  More recently, it appears the opposite approach is being taken.  Since ROS levels are already high in cancer cells, increasing the levels further may tip the balance and trigger cells death instead.  In normal cells, the response to ROS can vary depending on the levels.  Lower levels of ROS may lead to DNA damage which triggers growth arrest so that the cell can repair the damage.  Increasing ROS levels further may result in irreparable damage and so the cell may enter a state of senescence (permanent growth arrest).  Increasing ROS levels even further can cause so much damage that the cell undergoes apoptosis or even necrosis.  So a cancer cell that already has high levels of ROS may only need a little more to induce death.  Importantly, it is hoped that the levels of ROS needed to increase death in cancer cells is insufficient to cause death (or senescence for that matter) in healthy cells, which may be important for drugs that are non-specific.