LAST WEEK'S report of an "outbreak" of brain cancer at two Texas petrochemical plants is almost certain to trigger another round of fear, argument and confusion about the prevalence of occupationally caused cancer.

The last round began in 1978 when the government's top cancer experts published a study estimating that 20 to 38 percent of cancers are associated with occupational exposure. Since earlier estimates were in the range of 1 to 5 percent, the new study seemed to indicate a burgeoning epidemic of cancer in the work place, and it was generally reported in that light.

But the findings were soon criticized by an array of other experts as being much too high. Eventually, detailed comparisons between the government study and the others produced explanations for many of the differences. A more recent review pegged the fraction of cancers that are related to (not necessarily caused by) occupational exposure at 10 to 33 percent.

What these studies have clarified is the fact that because so many essential facts are still unknown, estimates of the occupational cancer rate are, by scientific standards, hardly more than gross guesses. For example, cancer incidence -- the rate at which new cases of cancer appear in the population at large -- is the basic number with which all others must be compared. Suprisingly, for the nation's second-largest killer, the rate is still uncertain.

Determining what proportion of cancers can be attributed to occupational exposure introduces many additional complications. The latency period of most cancers is between 20 and 30 years. Many of the most carcinogenic substances now being made were not produced that long ago, or were made in the past in small quantities. It is rare that records of workers' exposure are kept for so long a period of time. Presumably, there are also many substances now in the work place whose carcinogenicity has not been discovered.

Smoking, a contributor to many different cancers, further confuses the data.

The risk of lung cancer to an asbestos worker, for example, is 4 to 5 times as great as for a non-worker, but the risk to a worker who also smokes is about 40 times that for a non-worker, non-smoker. Yet it has also been reported that the lung cancer rate is rising among non-smokers. Other environmental contributors to cancer, including air pollution and diet, are equally hard to identify and measure.

All in all, statisticians and epidemiologists have an immense task, lasting years, before anyone can say with confidence whether there is or isn't an epidemic of cancer in the work place. In the meantime, Congress and the courts have to gain a much better understanding of the valid reasons for the continuing scientific uncertainty -- those listed here are only a small selection. Together with the Occupational Safety and Health Administration, they must then develop a clearer standard for regulating exposures to carcinogens where there is uncertainty about the degree of risk. Unless this is done, every OSHA standard will continue to undergo years of litigation, and hundreds of thousands of American workers will be needlessly and unfairly exposed to a terrible risk.