Despite allegations that police are “cooking the books,” the public should trust the city’s crime statistics, a New York University study says.

The study released this month analyzed the systems for monitoring the accuracy of police reports at the New York Police Department and eight other major departments. The researchers, an NYU professor, Dennis Smith, and a SUNY-Albany professor, Robert Purtell, concluded that “the City and department officials, and the public can be reasonably assured that NYPD data are accurate, complete, and reliable.”

They praised the department as having “by far the most robust and systematic quality-control approach in current use.”

Since the police department began using a statistical system called Comp-Stat to measure crime in 1994 under Commissioner William Bratton, there have been allegations that officers and their superiors were downgrading serious crimes to misdemeanors to make it appear that crime was dropping in the city. It was, in part, this perception that led Mr. Smith to begin the study.

“I was using the numbers myself,” Mr. Smith said. “I wanted to find out: How recently have these number been checked? Do we know how these numbers are checked?”

The study found that the ratio of petty larceny to grand larceny didn’t change much from year to year, which, Mr. Smith said, shows that there was no systemic downgrading of crime among police.

An officer or higher police official “cooking the books” would be likely to reduce a grand larceny to a petty larceny because the FBI monitors only the seven major felony crimes, not misdemeanors.

The police department has two bodies that monitor the inputs that create crime statistics, the deputy commissioner of strategic initiatives, Michael Farrell, said in an interview. Twice a year, the members of the Quality Assurance Division arrive unannounced at each precinct and pull a sample of 300 crime complaints. They compare the hand-written police reports with the official records, he said. They also call a selection of victims to find out whether the crime corresponds with their account, he said.

The Data Integrity Unit analyzes a larger volume of reports in the department’s computer system by comparing the crime listed with the narrative written out by the officer who took the report, Mr. Farrell said.

The researchers did not observe the actual auditors conducting the sampling or test the accuracy of the sampling process, but Mr. Smith said the guidelines were sound.

In 2000, the state comptroller, Carl McCall, completed a smaller study of the reliability of CompStat data, concluding: “City and Department officials can be reasonably assured that CompStat data are accurate, complete, and reliable.”

“This report is really to the police commissioner’s credit,” a professor of law and police science at John Jay College of Criminal Justice, Eugene O’Donnell, said. “It’s an urban legend that there is some systematic program of cooking the books.”