Gen. James Mattis thinks about when, and how, American troops should put their lives at risk.

Any risks—whether, for example, singing onstage, starting a company, or rock climbing —pale compared with the risks a soldier takes in combat. A soldier risks his own life, the lives of his comrades, and the lives of innocent civilians. An officer has this burden, and more, because he also makes the decision to risk the lives of his soldiers, knowing that some of them will come to harm.

Marine Gen. James Mattis, 59, has been making these decisions for almost 40 years since his graduation from Central Washington University. He led combat troops in the first Iraq invasion as a lieutenant colonel. He commanded Marines as a brigadier general in Afghanistan in 2001. In 2003 he was the Marines ground commander in Iraq, leading the 20,000 troops of the 1^st Marine Division for 500 miles over 17 days, the longest sustained march in Marine Corps history. He returned to Iraq months later to direct the fight against insurgents in the raging Al-Anbar province. Now a four-star, Mattis is commander of U.S. Joint Forces Command. It's his job to help the Army, Navy, Air Force, and Marines fight in coordination. He has also become a key figure in the debate over how the military should adapt to irregular warfare, the kind in which enemies hide in mosques or deploy computer viruses.

And Mattis has made a special study of risk. After returning from Iraq, he pushed to create the Marine Infantry Immersion Simulator. Built in an enormous former tomato-packing plant, the training course helps reduce the risk of friendly-fire accidents by re-creating the chaos of close-quarter combat. It also uses holograms to help Marines make the split-second decision between shooting an enemy turning the corner with a bomb and sparing the woman with a loaf of bread. In 2006, Mattis and Army Gen. David Petraeus led the push to write the Army/Marine Counterinsurgency Field Manual, which guides troops in Afghanistan and Iraq. The manual articulates a new concept of risk: Troops use less force and accept more short-term vulnerability to build ties with locals that will bring longer-term security.

Mattis is an evangelist for risk with two core principles. The first is that intellectual risk-taking will save the military bureaucracy from itself. Only by rewarding nonconformist innovators will the services develop solutions that match the threats conceived by an enemy that always adapts. The second is that technology cannot eliminate, and sometimes can't even reduce, risk. Mattis warns about the limitations of sophisticated weapons and communications. They can be seductive, luring military planners into forgetting war's unpredictable and risky nature, leaving troops vulnerable.

Fighting in Al-Anbar: Returning to History

If you've ever taken a big risk, this will sound familiar: As you approach the moment of decision, your heart spurs you on, but before you can act, your head gets in the way. You think, and you hesitate. You talk yourself out of starting that company or you lock up on the mountain you're climbing. Maybe you've drawn a line down the middle of a piece of paper and listed pros on one side and cons on the other. (Or, like Hamlet, you've wandered the house muttering.) Finding the balance between your head and your heart determines the difference between a risk and a gamble: With a risk, you can survive if you fail. With a gamble, a loss is irrevocable.

Mattis embodies the risk-taker's mix of head and heart. You can see it on the walls of his library. As one of nine combatant commanders, he was assigned the sprawling 17,000-square-foot Virginia House on Norfolk's huge Navy base. Unmarried, Mattis lives alone (Warrior Monk is one of his many nicknames). Walking into his pristine house I felt like I needed an admission ticket until I got to the two well-used rooms off a back hallway. The library shelves are packed with histories and military manuals. In conversation, Mattis regularly gets up to retrieve a volume—to cite a passage about the insurgency in Algiers or show a table about fuel use in the initial sprint into Iraq.

The photographs that hang on the wall space not devoted to books testify to the less cerebral side of Mattis' personality. One of his favorite photographs of many from his combat tours shows the men of the platoon he traveled with in Iraq. He did not command from a remote location as some generals do but made regular tours into the thick of the action. (In a five-month period in 2004, 17 of his platoon's 29 members were killed or wounded.) In another photograph, he's a young lieutenant in full combat gear, staring into the screaming mouth of his commanding officer. He is being chewed out for getting into a drunken bar fight the night before.

In Afghanistan, Mattis yelled at his driver for speeding through a village but complains that at home in the United States, Marine regulations force members of the corps to wear helmets when they ride motorcycles. He is not a barrel-chested Marine but of average build. In our discussions, he was easy to know and self-deprecating almost to a fault, attributing any battle success to the Marines and crediting most ideas to someone else. He was as much smart-ass as the hard-ass portrayed in HBO's Generation Kill. Greeting an officer in the Royal Navy, he quipped, "celebrating another year of not being French."

In the winter of 2003-04, the U.S. military needed any kind of solution in Al-Anbar province and turned to the Marines and Mattis. In the spring of 2003, Mattis, had led the 1^st Marine Division into Iraq to start the war. Now he was being called back into some of the ugliest fighting of the entire campaign, against a growing insurgency setting roadside bombs and hiding in Iraqi National Guard and police uniforms. This experience would test and shape his views about risk, the limits of force, and the necessity of adapting quickly in the fiercest conditions. In the end, the strategy he developed for Al-Anbar would require two things: lethal force and an intellectual theory of restraint.