Though the Street administration has made dramatic claims about the effectiveness of extra police details at hundreds of drug corners, police statistics show only a slight decrease in violent crime since the initiative began, an Inquirer analysis shows.

But residents are less likely to find their cars stolen or to report break-ins at their homes since Operation Safe Streets began in May. Property crime fell by 16 percent citywide from May through August.

Murders, rapes, robberies and aggravated assaults dropped 2 percent citywide during the first four months of Operation Safe Streets, from May to August, compared with the same months last year.

Violent crime was already down far more last year, before the antidrug initiative existed. From May to August 2001, police recorded 12 percent fewer violent offenses than they did during those months in 2000. Serious crime overall has been decreasing in Philadelphia since 1998.

Police and administration officials countered that crime has continued to fall since August - and they released data showing that a drop in murder has accelerated since the program began.

"I think based on our statistics and based on some of the community meetings, we have had a great impact," Police Commissioner Sylvester Johnson said.

Mayor Street declined to comment, referring all questions to Johnson.

There were 140 homicides in Philadelphia from May 1 to Oct. 24 this year, compared with 166 during the same period last year, a 16 percent reduction. But the impact of Safe Streets on murders is tough to pin down because last year, police received an unusually high number of reports after Sept. 11 when officers were redeployed to antiterrorism efforts.

Property crime, especially car thefts, did fall measurably during the first four months of Operation Safe Streets, the most recent time period for which official data were available. Burglaries, thefts and car thefts were down about 16 percent, compared with a drop of about 2 percent the year before.

Some of the numbers on violent crime are at odds with recent claims by Mayor Street, who has credited the operation with dramatically reducing "drug-related violence" by as much as 51 percent.

The mayor's reports of success have come as he is poised to ask City Council for more money to fund the multimillion-dollar program, although he has declined to divulge how much it has cost so far.

Police estimated this summer that they were spending more than $1 million a week on overtime for Safe Streets. The city has spent $17 million more for police overtime since May than it did during the same period last year, according to the City Controller's Office. How much of that was spent on Safe Streets is unclear.

In addition to city money, Street has poured immense political capital into Operation Safe Streets, appearing at neighborhood rallies across the city throughout the summer and fall touting its effects.

But like mayors before him, Street earlier this month learned the pitfalls of relying too heavily on crime data to bolster his political position. At a news conference in early October, he said drug-related murders had dropped from 51 to 9 between May 1 and mid-September and credited Safe Streets with saving 41 lives. Actually, Police Department data showed that murders classified as drug-related went up by one during that period - from 45 to 46.

Administration officials retracted the figures and then declined to release a corrected number of drug-related homicides, saying they were working on developing a new standard for defining which murders were linked to drugs.

In 1998, then-Mayor Ed Rendell was forced to retract crime statistics after claiming Philadelphia was the "safest big city in the nation."

Actually, it turned out, the police method of tracking statistics was unreliable and unscientific. The crime-counting system has since been reformed, but experts say it is difficult to get an accurate picture of crime trends even when the data are reliable.

University of Pennsylvania criminologist Lawrence Sherman warns that because crime trends can be caused by many different factors, the only way to be certain that a policing program is having an impact is to create "control" areas that are not included in the program and compare them with neighborhoods where police are active.

"You never know if the program had even more impact than you thought until you sort it out this way," Sherman said. "It's just as plausible that you'd underestimate the effect than that you'd overestimate it."

Police constantly update their figures. An aggravated assault could be classified retroactively as a homicide, for example, if a person who is shot and wounded dies months after the attack. And even the weather can affect murder rates.

"There are all sorts of fluctuations," said former Police Commissioner John F. Timoney, who was hired soon after the Rendell administration's statistical debacle because of his expertise in using up-to-the-minute data to pinpoint crime hot spots. "Even if you hold steady, some summers are hot. Some are cooler... . None of this is a perfect science. What you're looking for is, which way is it headed?"