A National Institutes of Health committee that had endorsed in principle abolition of federal regulation of gene engineering experiments yesterday backed away and settled for relaxing the rules.

The apparent change of heart came after the NIH Recombinant DNA Advisory Committee (RAC) received a batch of strong letters from university biosafety committees, congressmen and scientists, the majority of whom favored keeping mandatory federal controls on gene-splicing research.

The scientists' chief concern was that the absence of federal regulations might create a political vacuum that could result in more restrictive local laws and regulations. After nearly three hours of debate, the committee voted 16-to-5 to keep mandatory rules but relax their provisions.

Committee members were at pains to point out, and to incorporate in a preamble to the new federal rules, that keeping the mandatory regulations does not mean the committee believes gene-splicing experiments are more hazardous than other kinds of biological research.

The technology in question, created about nine years ago, allows the mixing of genes from two or more species. Biologists originally feared that putting genes from one kind of microbe into another might transform a harmless bug into a disease-causing one.

Yesterday several committee members said their chief fear was no longer scientific but political. They worried that a vacuum would be created where many differing and restrictive local laws might step in.

"There are legislators eager to use this as a political football," said RAC member Mark Saginor of the Metabolic Research Medical Group Inc. in Los Angeles. He said some legislators in California were ready to "whip up some scare tactics and get people behind them."

James O. Mason, director of the Utah Department of Health, said the federal rules "should and probably will become voluntary guidelines. But the time for that is not yet. The public is not yet ready."

David Baltimore of the Massachusetts Institute of Technology, who proposed the abandonment of mandatory rules, admitted there might be some problems with ending federal regulations because of "the fear of local regulations."

But he maintained that keeping the rules for political and not scientific reasons was "a sham . . . guidelines that are not respected in the eyes of the scientific community are very poor."

He said the NIH committee should establish national policy by declaring its opposition to mandatory rules, hoping that local agencies would follow.

Several committee members expressed worry over what they called the double standard being set by keeping rules on DNA research while allowing other similar and equally hazardous biological work to go without regulation.

King Holmes, head of the Division of Infectious Diseases in Seattle's Public Health Service Hospital, said the worries over gene-splicing work were really concerns about manipulating dangerous organisms.

He said two of the experiments most worrisome to the committee--giving an organism the genes to make biological poisons and giving an organism the genes to make it resistant to antibiotic drugs--could be done without using gene-splicing techniques.

Such experiments would be subject to no federal rules.

The current guidelines allow more than 90 percent of gene-splicing experiments to go on without special oversight. The revisions passed yesterday will expand this category to include essentially all non-disease-causing organisms.

The few experiments that require special review and approval by NIH or local biosafety committees are those that mix the genes of dangerous organisms.

Yesterday's actions must be approved by the NIH acting director before becoming final.