The National Institutes of Health formally declared yeaterday that there is a "small" risk, but a risk that "cannot be ignored," experiments in gentric engineering will set loose new germs or biological poisons to harm people.

It is also possible, NIH said, that methods "similar to or derived from" this research "may in the future" be used to deliberately modify "complex organisms, including humans," by altering hereditarp traits.

There is also, said NIH, a theoretical but unproved possibilty that someone might use the same techniques to make new agents for biological war or sabotage.

All three assessments were made as part of NIH's environmental impact statement on its safety guidelines, and in effect its support, for biology's newest advance: the "recombinant DNA" technology that joins genes of different organisms to make new or changed forms of life.

Though NIH listed many "possible hazardous situations," its conclusion - prepared under the direction of Dr. Donald Fredrickson, NIH director - was that it is important to do the research. The work is needed, NIH said, both to assess its safety and learn to control,it, and to learn its potential "for good and harm."

Because NIH both acknowledges possible harm and calls it purely "speculative," "theoretical" and unlikely, the statement will probably be used as ammunition by both friends and critics of the new studies.

The report, in fact, includes a few acid statements by critics. Dr. Erwin Chargaff, professor emeritus of bio-chemistry at Columbia University, warns that "you cannot recall a new form of life." The work is "bound to have evil results," he says, and "the future will curse us for it."

Dr. Richard Goldstein of Harvard University, in article reprinted in the report from the New England Journal of Medicine, says the research has "the potential for a new form of pollution, biologic pollution," and should proceed only on a limited basis, perhaps at a few isolated facilities.

Instead, as the statement shows, it is being done at several institutes of NIH in Bethesda and in some 200 NIH-funded projects at universities.

The report points out that the research was already going on in many places when in June, 1976, NIH issued guideline to govern it in NIH-funded institutions and to bar many experiments - for example, any to make bacteria or viruses more drug-resistant.

The statement acknowledges that the guidelines were issued before all possible effects were assessed and an impact statement issued under the National Environmental Policy Act. But if NIH had waited, the report says, research would have continued and the escape of potentially hazardous organisms would have been much likelier.

Virtually all the research in this country, some $10 million worth this year, is being funded by NIH. Some critics have said NIH could easily have declared a moratorium by halting funding.

But the statement also says research is proceeding in several nations, and scientists there were in fact waiting for NIH to set the lead in issuing safety rules.

Most of the research so far places genes from other organisms in a modified strain of E. coli. E. coli are bacteria that inhabit the human digestive tract, but it would be "virtually impossible" for the modified form to cause human disease, the statement said.

At the same time, it acknowledged that there are "few relevant experimental data" on what would happen if some future organisms, such as viruses, escape from laboratories, though the best estimate today is that the "level of risk" would still be "extremely small."

As to some larger issues - such as future modification of man - the statement does not even try to deal with them since, it points out, they are not the subjects of the current research. Such issues, it says, should be discussed by the advisory bodies that would be created by the recombinant DNA control bills now bogged down in Congress.

As to biological warfare with recombined DNA, the statement says it is prohibited by the Biological Weapons Convention to which the United States has subscribed.