The apparent problem is uncontrolled sectarian violence in Baghdad and the apparent solution is to send more American soldiers to restore order. The actual problem is a dysfunctional, sectarian Iraqi political system. Here at home, the imminent debate between the Congress and the administration about the number of American forces is a diversion. We may need more resources, but first we need a strategy.

President Bush faces a difficult strategic choice. First, he can continue to play defense and send in more troops to undertake tasks approved by Prime Minister Nouri al-Maliki. The intent is to buy another year during which a nonsectarian Iraqi government will pull itself together. It is difficult to see how it will. Alternatively, he can adopt an offensive strategy with clear benchmarks, strengthening Iraqi security forces while imprisoning Sunni insurgents and Shiite death squads. The risk is that Mr. Maliki may refuse to cooperate, forcing us to walk away. In sum, what lies before the administration is a final strategic choice, after a series of large and consequential failures.

The campaign began brilliantly with the swift rush to Baghdad in May 2003. Stripped of their dominance but otherwise untouched by the war, the Sunnis supported an insurgency led by Saddam loyalists and fueled by xenophobic religiosity. The American military stubbornly responded for 18 months with conventional sweeps and raids that fueled resentment and raised recruits for the insurgents. It was not until late 2004 that our military seriously began training a new Iraqi army. That same year, the White House endorsed a counterinsurgency strategy of "clear, hold and build" inside Sunni cities, despite Secretary of Defense Donald Rumsfeld's reservations that this thrust American soldiers into an occupying role without addressing how they would get out.

For a brief period, it appeared that "clear and hold" was working. Then, in February 2006, the Sunni extremists destroyed the Shiite mosque at Samara. This ignited the civil war they had sought to provoke for three years. As with the initial Sunni insurgency, our military was caught a second time without a counterstrategy.

Since the summer of 2006 we have fought a full-fledged, two-front war, waging one against the Sunni insurgents north and particularly west of Baghdad, and the other in Baghdad, where both Sunni and Shiite killers murdered the innocents. The Sunni tool was the massive car bomb. The Shiite militias were more systematic, employing death squads in a slow, methodical ethnic cleansing of neighborhoods.

The war seesaws back and forth. Unlike in Vietnam, these enemies do not dare to fight even small American military units. Our casualties come from roadside bombs and sniper attacks. In Anbar province, where the fighting is heaviest, our Marines are having unappreciated success in splitting local tribes from al Qaeda in Iraq. The under-strength Iraqi Army has performed well, despite grossly inadequate support from Mr. Maliki's government. The Iraqi police in Baghdad, though, remain penetrated by political militias. The spectacle of Saddam Hussein's guards chanting the name of Moqtada al Sadr as the despot swung from the gallows was revealing and disturbing.

So where do we go from here? The much-debated "surge" would only modify our current strategy, if Americans continue to focus chiefly on destroying the insurgents in Sunni neighborhoods. Fewer Sunni car bombs will supposedly result in less retaliation by death squads and end the ethnic cleansing. The Shiite population would turn away from militias because they wouldn't need them, given more security by more Americans. Because Mr. Maliki values Mr. Sadr's political support, Sadr City will remain what it is today: a sanctuary for the Mahdi Army militia.

This is a problematic approach. The Sunnis initiated the violence, no doubt about it. But now the Shiite militias are doing the majority of the killing in Baghdad -- yet less than 10% of those in prison are Shiites. There are 75 murders a day in Baghdad, and most killers walk free. The militias are gradually succeeding in the ethnic cleansing of much of Baghdad. To persist in this strategy is equivalent to a mayor telling his police chief that the mafia who live on the east side of town cannot be touched. It dooms the chance, however frail, of creating a nonsectarian Iraqi government.

Coupled with defensive patrolling is a proposed program of job creation (removing trash, for example). But handouts in a culture accustomed to handouts gain little. The return for this effort is hard to predict. In Anbar province, the only successful projects have been repayable loans to those Iraqis resolute enough to step forward and accept responsibility for building their own businesses.

Providing our advisers and battalion commanders with discretionary funds is a sound investment, but is ancillary to establishing security. It's unproven how many 16-year-old foot soldiers will be diverted by low-paying jobs. If you went to a member of the mafia and offered a low-paying job in return for renouncing crime, he would laugh at you. Most of the thugs won't be bought off; extortion and robbery are more lucrative and enjoyable.

President Bush recently agreed with the assessment of the chairman of the Joint Chiefs of Staff, Gen. Peter Pace, that "we're not winning in Iraq . . . but we're not losing." But in counterinsurgency, if you do not win, you lose. What, then, is another option the president can choose?

Instead of a defensive surge strategy satisfying to Mr. Maliki, the president can opt for an offensive, nonsectarian strategy. Its core operational concepts must be neutralizing criminals -- which include the Sunni insurgents, the Shiite death squads and the criminal gangs -- by imprisonment, deterrence, or death; and constructing Iraqi security institutions as free as possible of sectarian taint.

Iraq is now a police war and we need to treat it as such. Former Secretary of State Colin Powell a few weeks ago said, "We should not use our troops as policemen." But that's exactly how they are being used today in Baghdad and are, in fact, used in most counterinsurgencies. Our weakest links are leaving the Mahdi Army off-limits, not selecting Iraqi security leaders and refusing to arrest and incarcerate the criminals (insurgents, death squads and thugs). If the president's new strategy does not aggressively rectify these three defects, then surging more American troops will buy time but not alter a war we are losing because we are not winning.

Sadr City cannot remain off limits. When the death squads know they are hunted, many will flee the city. Others will fight back. Intense violence, however, cannot sustain itself. American forces fought Mr. Sadr's militia in April and August of 2004. In both cases, all-out war by the Mahdi Army petered out due to lack of logistics. In both cases, the Shiite population stood to one side. We created a monster by letting Mr. Sadr go free twice. We cannot make that mistake a third time.

To change the dynamic in Iraq, the president has to insist on arrests and incarceration rates equivalent or greater than those for violent crime in New York City. This would set benchmarks and shift our forces from defense to offense. New York City averages over 27,000 arrests a year for violent crimes. If a similar number of arrests were made in Baghdad, which is roughly the same population size as New York City, the jail population in Iraq would double in a year. One in 75 American males is in jail, compared to one in 450 Iraqi males. Iraq is not six times safer than the U.S. Instead, Iraq has a judicial system, abetted by American military legalisms, that works in favor of the killers. The president has to change that.

Our troops in Iraq complain, with justice, that they often capture insurgents, only to find them on the street a few months later. As a result of the abuses of Abu Ghraib, the U.S. military instituted four layers of review for each Iraqi detained. The result is that most detainees are released. As for the Iraqi system, it is simply absurd, insisting on habeas corpus rules of evidence in a corrupt and overwhelmed judicial system that incarcerates a few dozen each week, compared to over a thousand a week in New York City alone. Eight of 10 detainees are set free. Releasing killers undercuts troop morale, while the residents lose trust. Texas has 170,000 in jail; Iraq, with a larger population and 50 times the violence rate, has 28,000 in jail. This "catch and release system," as the troops call it, is the single weakest link in the U.S. strategy.

Technology can help. We could, for example, equip Iraqi and American forces with mobile devices to fingerprint the military-age males in Baghdad and the Sunni Triangle. Anyone stopped can be checked in two minutes, just as the Border Police and the Chicago Police do today. This would deprive the killers of mobility and is the key to radically increasing arrests. In this, as in many other respects, the American failure in Iraq reflects not our preference for high technology -- as facile critics claim -- but our inability to bring appropriate technologies to bear.

This is an Iraqi war, and success depends on the creation of a larger Iraqi Army (perhaps twice its current size of under 140,000) and a neutral police force. So we must increase our advisers from 3,500 to 15,000 or more. This is a small-unit police war, with the insurgents hiding and dodging. We do not need 40 or 50 conventional American battalions trained and equipped for full-scale conventional war, if the Iraqi security forces are strengthened by bulked-up American advisory teams.

But Iraqi security forces will fall apart if political parties use appointments and promotions for sectarian purposes. We must therefore insist on a joint U.S.-Iraqi board to appoint all Iraqi battalion commanders and police chiefs and above. American control over senior personnel is the single most important aspect of our effort to develop Iraqi institutions. Without it, additional resources, to include more advisers, will be wasted.

Quite possibly, Prime Minister Maliki will refuse, on the grounds of sovereignty and national pride, to allow Americans equal control over Iraqi personnel policy. We should respond that when Iraq is truly sovereign and standing on its own, we withdraw our advisers and the joint board ceases to operate. In the meantime, we're not potted plants. It is our advisers that force the ministries in Baghdad to pay the Iraqi soldiers. It is our advisers on patrol risking their lives and dying to reassure the Iraqi forces that they can prevail. As long as we run equal risk, we deserve equal say in the selection of competent leaders.

There is some chance that a strategy dependent on Mr. Maliki's sectarian instincts may succeed. He may, despite his poor track record, pull himself and his government together. Prudence, however, suggests that the president design a strategy that is independent of Mr. Maliki's fortunes and strengthens the only institution capable of holding the country together -- the Iraqi army.

We prefer an offensive strategy based on three ironclad principles: take the offense immediately against the death squads in Sadr City, who are now unsettled; arrest and imprison on a scale equal to the horrific situation (or at least equal to New York City!); and insist on a joint say in the appointment of army and police leaders. If the Iraqi government refuses, we should be willing to disengage completely, and soon.

The paradox of American strategy in Iraq is this: President Bush can achieve success only by threatening to do something he is morally opposed to doing -- leaving swiftly and risking chaotic civil strife. If the president showed the same iron will toward Mr. Maliki that he does toward Congress and public opinion, Mr. Maliki would blink first.

The only course that will work entails not only the risks of greater casualties, but the risks of walking away from promises unmet and hopes unfulfilled. More money and troops are inputs, not outputs. A new strategy needs benchmarks for success -- arrests, imprisonment and the adviser ratings of leaders. Our only hope lies not in American troops but in the development of an Iraqi security force free of militia influence, working for a government that understands the penalty for failure.

Mr. West, a former assistant secretary of defense and a former marine, visits frequently with the troops in Iraq. Mr. Cohen is Robert E. Osgood professor of strategic studies at Johns Hopkins School of Advanced International Studies.