Rick Medina sat down at a small microfiche reader at about 7:30 Monday morning, inserted a strip of film, then started pushing buttons on a desk calculator. Fifteen minutes and 15 calculations later, the 44-year-old Bureau of the Census statistician became the first person to know if the national unemployment rate hit the landmark 10 percent level in September. Once he knew the answer, Medina had to figure out how to forget about it for the next four days.

Since Monday, the figure that could be a weighty albatross dragging down the stock market and the hopes of GOP congressional candidates, the figure that could mean quick profits for canny investors in the bond market, has been the private province of about 15 statisticians and managers in the census offices in Suitland and the Bureau of Labor Statistics office in downtown Washington.

The security surrounding the data is unpretentious, consisting largely of closed mouths, locked file cabinets, a few private computer access codes and one small "No Admittance" sign on a drab blue-green door. But these statisticians can remember no more than two or three cases in the last 15 years when the numbers leaked out.

"Everybody knows not to discuss this outside the office under any circumstances," says Jack Bregger of BLS. "We don't even talk about these things to our spouses . . . particularly at sensitive times like this, with unemployment rising, and particularly just before an election."

The secrecy begins with about 130,000 people in 60,000 American households whose lives are the stuff of which the unemployment statistics are made. They are guaranteed confidentiality by the Census Bureau, though it's up to them if they want to tell their neighbors what they're doing.

The secrecy extends to 1,500 part-time Census Bureau employes who drop in or telephone the 60,000 households once a month to ask predetermined questions and record the answers on machine-readable forms: who has a job, what industry they work in, who's been laid off, how long, and so on.

The interviews started Sept. 20. By Sept. 22 the forms were coming in to 12 regional census offices, which forwarded the figures to the bureau's data processing center in Jeffersonville, Ind. By Sept. 30, all the data had been transferred onto one master computer tape in the Suitland offices of the Census Bureau. There, over the weekend, the figures were massaged, weighted and made ready for the calculators at the Bureau of Labor Statistics.

On Monday morning, Rick Medina and his bosses, Ken Riccini and Greg Russell, came in to the census offices to punch the calculators one more time.

"There's definitely that feeling of anticipation -- of 'What's it going to be?' " said Russell, who has sat before the microfiche readers and the calculator many times in the past. "It's a feeling of curiosity before, and when you know, it's a feeling of uniqueness." Every month, on a rotation basis, one of the eight statisticians in Russell's current surveys office does the honors.

For Medina, a recent transfer from the offices of the Decennial Census, this was all new. "I just wanted to be sure I got it right," he said.

The figure is important in any month, but the September one -- the last to be released before the November elections -- has been the object of intense speculation because it could break the double-digit barrier. "If it's over 10 percent," says Rep. Tony Coelho of California, chairman of the Democratic Congressional Campaign Committee, "that'll be the dominant issue from now till Election Day."

When Medina finished his calculations, he boxed the relevant material, getting it ready to hand over to John Stinson, who would take it back to BLS. In the meantime, Bregger called Riccini, who gave him the 12 basic figures that must be seasonally adjusted and combined to produce the unemployment figure.

Bregger sat down at his desk calculator and made the same 15 calculations Medina had made, adjusting the figures for seasonal variations in unemployment, checking his results against those of the Census Bureau. Then he dropped a copy of the figures off at the office of his boss, Janet L. Norwood, commissioner of Labor Statistics, giving her the news that the rest of the world won't get until 8:30 this morning.

Between Monday and yesterday, thousands of calculations were made. The tables breaking down employment and unemployment into categories: agricultural employment, employment by industry, employment among Vietnam-era veterans -- 651 separate statistics in all -- were sent to the print shop at the Frances Perkins Building.

By that time, Norwood had given the numbers to Martin Feldstein, chairman of the president's Council of Economic Advisers, who would break the news to President Reagan.

This morning, she was scheduled to brief Labor Secretary Raymond J. Donovan; copies of the figures were to be left in the Labor building lobby for Paul Volcker, chairman of the Federal Reserve, and Treasury Secretary Donald T. Regan.

More Cabinet members used to get early word of the data, but the distribution list was cut back in the mid-1970s after some leaks, Bregger said. The time of the release was also moved up from 10 a.m. to 8:30 -- before the financial markets open -- after a Chicago brokerage house complained that one of their competitors had a 15-minute head start on the news. Bregger believes one leak came from a radio reporter who called a broker for comment before the official release time.