WHAT ARE the causes of cancer? Recently a new set of answers has begun to emerge, still only in broad outline, but markedly different from earlier theories. The largest contributors to cancer deaths in the United States are now thought to be within the control of each individual, namely, tobacco, diet and a collection of behavioral and cultural actors often summarized as "life style."

At various times in the past, attention has focused on genetic inheritance, viruses, pollution and occupational exposures, all of them largely beyond individual control. Occupational exposure, in particular, is the source of a good deal of confusion. A 1978 study that seemed to represent the views of the government's principal health research agencies concluded that 20 to 38 percent of American cancers are work-related. Though that conclusion was widely criticized and swiftly repudiated by several of the report's authors, the perception remains that work-place exposures account for a large and growing fraction--some have called it an emerging epidemic--of American cancers. Yet scientists believe that occupation accounts for "only" (more on that n a moment) about 4 percent of cancer deaths. Similarly, pollution is thought to account for fewer than 5 percent of fatal cancers, medicines and medical procedures for about 1 percent, and food additives for another 1 percent.

On the other hand, tobacco alone is thought to cause 30 percent of cancer deaths and alcohol about 3 percent. Reproductive and sexual behavior, including such things as the age of first pregnancy and determinants of hormone balances, is estimated to cause about 7 percent of total deaths, though the known effects are considerably greater among women.

A major recent study by Sir Richard Doll and Richard Peto, two Britons considered by many to be the world's leading experts in this field, attributes 35 percent of cancer deaths to diet, including in that term the ingestion of carcinogens, the whole range of nutritional factors--fat, fiber, vitamins, trace elements--and perhaps also nutrients that inhibit rather than promote carcinogenesis. Drs. Doll and Peto emphasize that most of these dietary factors are still unidentified and that the total must be considered "speculative." Yet it conforms with a number of other estimates and is certain to have a major impact on medical thinking and research priorities.

None of these figures should be taken as firm. Some may turn out to be quite wrong. But one broad conclusion leaps out from the numbers: so far as is now known, many more cancer deaths seem to be due to avoidable, individually controlled behavior, than to involuntary exposures or inescapable consequences of modern living. This does not mean that identifying sources of occupationally caused cancer, for example, is any less important. Such cancers may have tragic impacts on certain small segments of the population, and once identified can generally be prevented. But it does have obvious implications for U.S. poliies. If these new estimates are approximately correct, past emphases in research, prevention and regulatory control may not have been doing the greatest good for the greatest number.