In her "President's Desk" column in the August issue of OR/MS Today, INFORMS President Susan Albin quotes Seth Bonder's keynote address at the 2007 INFORMS meeting, describing how the National Academy of Engineering defines operations research: "Development and use of analytical methods to describe, analyze, plan, design, manage and integrate the operations of systems and enterprises that involve complex interactions among people, processes, materials, equipment, information, organizations and facilities to provide services and produce goods for society."
The question of how to define OR crops up periodically in a variety of forums. Here's a thread on sci.op-research from 1999 about it, and here's a discussion on OR-Exchange from earlier this year. INFORMS Online has a brief definition on its home page (upper right) with links to delve a bit deeper (including, handily, separate definitions of OR and MS).
What struck me about Susan Albin's column was her interpretation of the NAE definition, and how it distinguishes OR from other engineering disciplines: "Each of the latter focus on physical processes ... In contrast, operations researchers focus on operations processes ..." So if we adopt the NAE definition (or perhaps a more compact version thereof? my fingers are still cramped from retyping it), all we need do now is figure out how to distinguish OR from operations management. I suspect the definition I use locally ("we do the same things, except we understand the math") won't fly globally.