A Justice Department unit is suspending a door-to-door survey of crime victims that has cost $6.5 million annually and has been heralded as providing "a whole new dimension" in assessing national lawlessness.

After collecting and analyzing crime victim data for nearly five years at a cost of $53 million, the program's managers yesterday cited budget shortages and questions about the survey's validity and reliability as reasons for the suspension.

But James M.H. Gregg, acting administrator of the Law Enforcement Assistance Administration, which has financed the project, said he thought that the survey would be restarted once major kinks are ironed out - "unless at some point we lose funding support. There's no indication of that," Gregg said in an interview.

A related survey on crime experienced by commercial establishments, which cost about $3 million annually is being totally scrapped. The two crimes covered by the commercial survey - robbery and burglary - are already relatively well reported to police, Gregg said.

Initiated in 1972, the national crime survey was designed to provide potentially crucial information on victims of crime and to measure the tremendous volume of crime known to be going unreported.

The FBI's Uniform Crime Reports - the only other national crime survey - reflect only crimes reported to state and local police. The LEAA survey found that actual crimes run anywhere from two to five times as high as reported crime in the nation's largest cities.

The Census Bureau conducted personal interviews for the LEAA every six months in some 60,000 households and 39,000 businesses, asking whether rape, robbery, assault, personal and household larceny burglary and auto theft had been experienced there.

But an assessment of the survey by the National Academy of Sciences, conducted last year at LEAA's request, uncovered what Gregg called "serious methodological problems" in the survey.