A requirement arose during decommissioning work at a UK Magnox Nuclear Power Station to identify the hazards involved in removing High Dose Rate Items from a Cartridge Cooling Pond. Removing objects from the cooling pond under normal situations is a routine event with well understood risks. The situation described in this paper is not a routine event. The power station has shifted from an operational phase in its life-cycle to a decommissioning phase wherein the risks, and procedures to deal with them, have become more uncertain. Are the hazard identification methods that have proven useful in one phase of the system lifecycle just as useful in another, and if not, what methods should be used? In this paper we will present an agenda for a contingent approach to method selection based on concepts from complexity theory and show the outcomes via two method case studies. Fundamentally, hazard identification methods need to be matched to important (though often not recognised) aspects of the real-life problems to which they are applied.