Tradesurveys have estimated that from 30% to 65% of IS development projects become 'runaways' --projects that 1) fail to produce an acceptable system, 2) grossly exceed initial budget/time estimates and, 3) seem to take on a life of their own. iven that the United States spends over $250 billion annually on IS development [Johnson, 1995], it is important to ask why runaway projects occur. In this research study, we focus on why such poorly performing projects are allowed to continue for so long. Based on datafrom several specific cases of runaway projects, we believe runaways are a form of organizational decision making failure. Consider this scenario: a development project encounters some problems which may be serious enough to cause the project to fail. Thedecision maker(s) responsible for the project have two choices: continue the project or abandon it (either terminate it or radically redirect it). If the decision maker consciously chooses to continue committing resources to the project or never consciously considers abandonment, the behavior represents an escalation of commitment or escalation, for short. Runaway projects result from many such decision points. Eventually these continuation (escalation) decisions take the project beyond its initial time/cost targets. Each subsequent escalation decision leaves these targets further and further behind until there is a feeling that 'the project is out of control'. Runaway projects, therefore, come from project escalation --the continued allocation of resources to projects despite negative feedback relating to project performance and the likelihood of success