Cancers are an example of evolution in action - evolution working not at the level of reproductive fitness of the organism but, instead, at the reproductive fitness of the individual somatic cell. Cancers arise when somatic cells accrete mutations that corrupt the functions and control of those genes that normally regulate, coordinate and curtail untoward cell replication, leading to clones of cells that expand and spread uncontrollably. To understand how cancers arise, we need to know how these constraints on untoward replication of normal somatic cells operate - how they sense normal versus abnormal cell behaviour and what they then do to reign in rogue cells. To selectively kill cancer cells we need to know which, of the plethora of random mutations that accumulate in cancer cells, are ‘mission critical' to cancer cell survival and propagation - yet at the same time are dispensable for the survival and proliferation of normal cells. And since cancer therapy acts as just another selective pressure, we also need to identify those drug targets whose functions are not only essential but also functionally non-redundant. In this way we can circumvent the principal reason for cancer drug failure - the emergence of resistant secondary tumours. In this symposium, we will explore these various issues - the tortuous and capricious way that cancers evolve and what this implies for cancer therapy, how we can identify which mutations are ‘mission critical' to tumour maintenance versus those that are merely noise or bit-part players, and how evolutionary concepts can guide us towards more specific and more durable cancer therapies.