Mayors bristle and crime experts roll their eyes when the ranking of "most dangerous" big cities are publicized each fall.

St. Louis typically finishes near - or at - the top, along with the likes of Detroit and Atlanta. This year it finished third in the annual report by Morgan Quitno Press, a private research firm in Lawrence, Kan.

But a new study of murder rates by a team of academics takes a swipe at the Morgan Quitno rankings and gives St. Louis and its beleaguered peers something to cheer about. Well, kind of.

St. Louis had fewer murders in 2004 than the study's researchers expected given the city's high poverty and other indicators of economic distress.

In a choice that might surprise some, the team listed San Francisco No. 1 in murder among the 67 biggest U.S. cities in 2004, despite a murder rate that was one-third of what St. Louis had.

St. Louis, which had the nation's fifth-highest murder rate, ranked No. 19 in the study. Detroit was 37th and Atlanta 46th.

San Francisco's mayor's office did not comment on the findings.

Why the poor finish in the Bay City? Because its 88 murders last year were more than researchers expected, given the city's relative wealth and low unemployment, said Richard Rosenfeld, a criminologist at the University of Missouri at St. Louis.

He co-authored the Improving Crime Data study with Robert Friedmann at Georgia State University, and Alfred Blumstein at Carnegie Mellon University. The work was funded by the National Institute of Justice.

The study predicts higher murder rates in cities that have more poor people, more blacks, higher divorce rates, higher unemployment, lower income and more single mothers. Blacks are victimized at a higher rate than whites, the researchers said, and cities like St. Louis with large black populations will experience more murders than cities like San Francisco with a smaller proportion.

Social and demographic factors drive roughly 70 percent of the crime rate and are generally beyond police control, the researchers believe.

By zeroing those out, the researchers said they are shining a light on other factors, such as the quality of a city's policing, that can be addressed. The study points to cities such as Cleveland, where murder rates were far lower than expected, whose policing strategies could be emulated.

The study used similar sets of data, but reached conclusions that differed greatly from the Morgan Quitno's more publicized rankings.

Morgan Quitno's assessment is based on a given city's rates last year in six crime categories - murder, rape, robbery, aggravated assault, burglary and auto theft - as reported to the FBI. The firm scored cities against national averages in each category and added the scores, weighing each crime the same.

The methodology has "no merit and no validity," said Friedmann. "And yet it captures headlines."

Scott Morgan, the Morgan Quitno president, defended his report Tuesday, saying he didn't want to criticize the other study, which he said seemed "awfully academic" compared to his.

Many people use his company's research to assess the relative safety of the place they live, he said. Morgan said he is tickled that university-based criminologists were ranking cities, a tactic he said they normally criticize him for using.

"It's nice that they acknowledge that it is of interest to people," he said. He said suggesting that certain groups, such as poor people, were more responsible for crime would be a "dangerous bit of categorization that I would want to avoid."

St. Louis City Hall still bristles when the city is compared to others, saying police are making strides in reducing crimes in city neighborhoods. But the new study was received more warmly than Morgan Quitno's.

"If you were to rank cities, this is the way to do it," said Jeff Rainford, chief of staff for St. Louis Mayor Francis Slay. The study is encouraging news for the police, he said, but shows the city has to deal with its concentration of poor.