Cigarette smoking, by contributing to heart and lung diseases, cancers, fires and other causes of mortality, is responsible for almost 16 percent of all deaths in the United States each year, according to an analysis by the federal Office on Smoking and Health to be published today.

The federal study, using figures for U.S. deaths in 1984 and scientific data on the contribution of smoking to various diseases, concluded that more than 320,000 of the approximately 2 million deaths in that year could be attributed to smoking, the chief preventable cause of death in the United States. The report appears in today's issue of the Centers for Disease Control's weekly bulletin.

The estimates are conservative, said Dr. Thomas Novotny, the government epidemiologist who prepared the study, because U.S. smoking rates in 1985 were used to make the calculations. The proportion of Americans who smoked was greater in previous years, so smoking probably contributed to even more of the 1984 deaths than the figures suggest.

"It reflects smoking over a continuum," Novotny said. "These deaths were related to smoking prevalence in the last three decades. You can't name every one of those people, but this is an estimate of the quantity of death due to smoking . . . ."

The report lists estimated numbers of deaths attributable to smoking from 24 different diseases, including heart disease, lung cancer and other cancers, strokes, pneumonia and chronic lung disorders.

It also estimated that more than 2,500 deaths of infants under one-year-old could be attributed to smoking by the mother. Various studies have shown that smoking during pregnancy increases the frequency of low-birthweight infants, premature births, lung disorders in the newborn period and sudden infant death syndrome (SIDS), Novotny said.

The total number of deaths attributable to smoking, 320,515, is higher than that estimated in previous studies, Novotny said, reflecting the fact that the recent mortality figures show the health impact of smoking habits over the last several decades. In 1980, smoking was estimated to have caused 270,000 deaths and in 1982, 314,000 deaths.

Americans' cigarette smoking rates have been declining since the 1960s, when reports of the adverse health effects were first widely reported. In 1986, 26.5 percent of adult Americans smoked, the lowest proportion ever recorded. In 1964, the year the U.S. surgeon general first reported that smoking increased the risk of lung cancer, 40 percent of Americans smoked.

Eventually, annual mortality figures from smoking-related diseases should begin to reflect Americans' declining smoking rate, Novotny said. He noted that death rates from heart disease are already decreasing and that lung cancer death rates in men appear to be leveling off. However, lung cancer rates among women are rising and the disease surpassed breast cancer in 1985 as the most common cause of cancer deaths in women, he said.

Novotny said that in compiling the figures, he only considered diseases for which a number of scientific studies had shown that smoking clearly increased the risk. "There are many other diseases one could think about" that were not included because their relationship to smoking was less definite, he said.

To estimate smoking's contribution to premature deaths, the researchers calculated that Americans dying of smoking-related causes before age 65 lost a total of 949,924 years of potential life.

The report said that smoking's contribution to the mortality rate for men was twice as great as that for women, reflecting the fact that a greater proportion of men than women were smokers over the past few decades.

Smoking's contribution to the mortality rate for blacks was 20 percent greater than for whites, and the number of years of potential life lost by blacks who died before age 65 was more than twice the figure for whites. The report said this difference reflected both higher smoking rates among blacks and higher mortality rates from diseases caused by smoking, such as heart disease and lung cancer.

Brennan Moran, assistant to the president of the Tobacco Institute, questioned the significance of the new estimates. "They've used statistics to come up with statistics," she said. "Whether or not there are biological causes that relate back to statistics is an open question."

The institute, a lobbying organization for the tobacco industry, takes the position that the role of cigarette smoking in causing disease is not scientifically established.