To examine the complexity aspect of postponement and its impact on supply chain risk management, we draw insights emerging from the theoretical principles in normal accident theory (NAT), where accidents in complex systems are presumed to be ‘normal’ or inevitable. NAT was originally developed to explain why marine transport is less prone to accidents than other systems like nuclear plants. However, this organisational theory has wider implications for understanding the structural characteristics of a system and then analysing the propensity of the system accidents and failures. In the recent supply chain management literature, Wagner and Bode (2006) suggest that NAT supports the presumption that modern supply chain management initiatives amplify the fragility of supply chains. They also call for future research on ‘the identification of further underlying factors (that drive supply chain vulnerability) by exploiting theories that yield exploratory power on supply chain vulnerability such as normal accident theory .. .’. NAT states that organisations may be prone to accidents under conditions of high interactive complexity and tight coupling (Perrow 1999a). Here, an accident is defined as a failure in a subsystem or system that damages more than one unit and in doing so disrupts the ongoing output of the system (Perrow 1999a). Coupling refers to the level of slack or buffer within the system while interactive complexity refers to the way that parts within a system are connected and interact (Sagan 1993). According to NAT, the likelihood of accidents can be reduced by making structural changes to reduce interactive complexity, or reduce tight coupling (Perrow 1999a). However, Perrow (1999b) points out that doing both at the same time will most likely make the system inefficient. Therefore, here we only consider the following two options companies are faced with to mitigate against disruptions: reduce tight coupling while remaining complex; or reduce interactive complexity while remaining tight coupling. Finally, NAT warns against the potential negative consequences of adding redundancy to a system such as increasing interactive complexity (Sagan 1993, Perrow 1999a, 1999b).