ETROIT -- What makes one automobile safer than another?

This used to be a simple question. Regulators looked at collision data. If more people tended to die while riding in Vehicle X than in Vehicle Y, then Vehicle Y was safer. In the 1970's, the government added crash testing to study how cars did in different kinds of collisions.

But now regulators and safety advocates are saying that another factor should be taken into account: how much damage does Vehicle Y inflict on Vehicle X when they collide? This "compatibility question," as it is called, is at the heart of the debate over the safety of sport utility vehicles, the American auto industry's most profitable product.

Last Wednesday, before a Senate Commerce Committee hearing on S.U.V. safety, the industry's top lobbying group held a pre-emptive news conference on the subject. "S.U.V.'s are as safe as cars and in some cases safer than cars," said Josephine S. Cooper, the president and chief executive of the Alliance of Automobile Manufacturers.

Her group presented data showing that 16.25 occupants of every 100,000 S.U.V.'s died in collisions annually, slightly higher than the 15.7 occupants per 100,000 passenger cars. Though many drivers bought S.U.V.'s believing they were much safer than cars because of their size, their high risk of rollovers, a particularly lethal kind of accident, more than offsets any advantage.

But critics say the true safety difference between S.U.V.'s and cars is much greater, if one considers what S.U.V.'s do to other vehicles. "It's really unethical to talk about the deaths only to the occupants of these vehicles," said Joan Claybrook, president of Public Citizen, a consumer group, at the Senate hearing. Ms. Claybrook was the top safety regulator during the Carter administration, when crash testing was begun. "You also have to look at deaths they cause to the occupants of other vehicles," she said.

In fact, some S.U.V. critics contend the auto industry uses a "kill or be killed" marketing strategy to sell these vehicles: if you don't want to be killed by an S.U.V., you'd better buy one yourself.

Two weeks ago, under unexpectedly intense pressure from the Bush administration's top auto regulator, Dr. Jeffrey W. Runge, the industry agreed to cooperate in an effort to develop voluntary standards to make S.U.V.'s and pickups less dangerous to cars in a collision. But last week, the industry said that the plan was no reflection on the relative merits of S.U.V.'s.

"This is something we're looking at because it's a whole new field of research," said Eron Shosteck, a spokesman for the Alliance of Automobile Manufacturers. "It's not just S.U.V.'s. It's about managing to protect the occupants of both vehicles."

Robert Strassburger, the alliance's top safety expert, dismissed compatibility as an issue that has been around "since the horse and buggy."

Still, it has drawn increasing scrutiny from regulators as the nation's mix of vehicles has changed. Twenty years ago, a fifth of sales went to light trucks (S.U.V.'s, minivans and pickups). Last year, they accounted for 52 percent of sales.

The future of G.M., Ford and Chrysler now depends on S.U.V.'s and pickups. The bigger models are exempted from many safety and environmental regulations, and under the Bush administration's tax plan, small-business owners could deduct their entire purchase price.

But whatever the corporate significance of large passenger vehicles, their effect on fatality rates is also an ethical issue, said Dr. Runge of the National Highway Traffic Safety Administration.

"The theory that I'm going to protect myself and my family even if it costs other people's lives has been the operative incentive for the design of these vehicles, and that's just wrong," he said in a December interview. "Not to sound like a politician, but that's not compassionate conservatism."

At last week's Senate hearing, he said that when an S.U.V. strikes a car from the side, it is three times more likely to cause a fatality than another car. Regulators believe this is mostly because of the high ground clearance of S.U.V.'s, not their weight.

Such data is reshaping the way scientists evaluate vehicle safety. Marc Ross, a University of Michigan physicist, and Tom Wenzel, a researcher at the Lawrence Berkeley National Laboratory, published a study last year that formulated a combined risk for different vehicles, based on the deaths they inflicted both on their own occupants and on others.

Pickups, subcompact cars and S.U.V.'s performed the worst, with the Chevrolet S-10 pickup the worst over all. Large and midsize cars, and minivans performed the best, with the Toyota Avalon sedan the best over all.

"A shortcoming of many safety analyses has been that only risks to drivers or occupants of a given kind of vehicle are evaluated," the study says. "Risks imposed on others are ignored."

Whether the matter of compatibility will be left for the auto industry to resolve remains to be seen. Dr. Runge, having secured the carmakers' cooperation, testified Wednesday that he preferred not to regulate.

But Senator John McCain, Republican of Arizona and the Commerce Committee chairman, said, "What's the credibility of the auto manufacturers when they clearly opposed seat belts and air bags?"