During an assessment, agents may use a limited set of techniques, including searching databases about targets, conducting surveillance of their movements and sending a confidential informant to an organization’s meetings. But to use more intrusive techniques, like secretly reading e-mail, agents must open a more traditional “preliminary” or “full” investigation. Such inquiries require agents to first have a greater reason to start scrutinizing someone: either an “information or allegation” or an “articulable factual basis” indicating possible wrongdoing.

According to the data, during the 2009-11 period agents opened 42,888 assessments of people or groups to see whether they were terrorists or spies. A database search in May 2011 showed that 41,056 of the assessments had been closed. Information gathered by agents during those assessments had led to 1,986 preliminary or full investigations.

The data also showed that agents initiated 39,437 assessments of people or groups to see whether they were engaged in ordinary crime. Of those, 36,044 had been closed, while 1,329 preliminary or full investigations had been opened based on the information gathered.

Michael German, a former F.B.I. agent who is now with the American Civil Liberties Union, said the high number of assessments relative to the number that developed into more intensive investigations was cause for concern. He noted that the F.B.I. retained the data it collects about a target, even if the person or group turns out to be innocent.

Newsletter Sign Up Continue reading the main story Please verify you're not a robot by clicking the box. Invalid email address. Please re-enter. You must select a newsletter to subscribe to. Sign Up You will receive emails containing news content , updates and promotions from The New York Times. You may opt-out at any time. You agree to receive occasional updates and special offers for The New York Times's products and services. Thank you for subscribing. An error has occurred. Please try again later. View all New York Times newsletters.

“It’s clear the F.B.I. is casting its investigative net too broadly,” Mr. German said. “And remember that only a small proportion of ‘preliminary’ investigations become ‘full’ investigations, and only a small percentage of full investigations result in criminal charges. This data makes it clear that the ‘assessment’ authority granted in the attorney general guidelines is far too broad.”

But Valerie E. Caproni, the F.B.I. general counsel, said that the data showed that agents had been able to dispose of about 96 percent of the low-grade reasons they might have had for suspecting someone of wrongdoing, like a vague tip or some other ambiguous lead, using “low intrusion techniques” rather than by opening a potentially more invasive preliminary investigation.

The new investigation standards, Ms. Caproni said, “end up being privacy protective because previously, without a well-developed, robust assessment category, many if not most of those would have been opened as preliminary investigations.”

The newly disclosed data roughly matched a far more limited disclosure earlier this year about assessments of people and groups generated in late 2008 and early 2009, but the latest data covered a longer period and included additional detail. The data also bolsters the F.B.I.’s assertion that since the Sept. 11 attacks, the bureau has taken low-grade tips about national security threats more seriously than similarly vague and seemingly implausible leads about possible criminal activity. The 39,437 criminal assessments were based on 73,303 complaints received by the F.B.I., indicating that about half of the time, agents merely filled out a complaint form but saw nothing worth following up on.

Advertisement Continue reading the main story

In contrast, the disclosure did not offer a separate number of national security complaints, suggesting that some were not followed up with an assessment. The bureau says its policy calls for every national security-related tip, no matter how dubious, to be investigated. Still, Ms. Caproni cautioned that the assessment data did not offer a complete view of the F.B.I.’s activities. For example, she said, if the Central Intelligence Agency told the bureau that an overseas source had provided a specific claim about a terrorist cell operating inside the United States, agents would immediately open a more intensive investigation rather than starting with an assessment.

She also cautioned that some details about the numbers might be fuzzy. In the past, for example, the F.B.I. has noted that one assessment may uncover information that leads to more than one preliminary investigation. At the same time, agents may occasionally have neglected, when opening a preliminary investigation, to record the fact that it developed out of an assessment.

The new data also showed that while the agency has opened fewer assessments looking for broader intelligence about possible threats and vulnerabilities within a district, a category it calls Type 3 assessments, it has been slower to close them. Of the 1,819 it had opened during the period, 1,056 were open by May.

Those figures covered Type 3 assessments generated both by national security programs — like domestic and international terrorism, weapons of mass destruction and counterintelligence — and criminal programs, like civil rights, gangs, organized crime, violent crime and white-collar crime. The F.B.I. censored the specific breakdown for each category. The data release comes as the F.B.I. is preparing to issue agents a new version of its manual, the Domestic Investigations and Operations Guide, that will relax some rules about the techniques allowed at the assessment stage. Ms. Caproni said the new rulebook is at the printer now, and a partly redacted version is likely to be released to the public at the same time it takes effect around mid-October.