Traffic deaths, which have remained below 54,000 a year for more than a decade, could shoot up to 70,000 in 1990 if there are no major new safety improvements in vehicles or highway design and no significant increase in seat-belt use, a Transportation Department study concludes.

Ray Peck Jr., head of the National Highway Traffic Safety Administration in DOT is expected to release the report Monday, following a speech to the National Safety Council in Chicago, an administration spokesman said. The report was made available by sources in the highway safety lobby who oppose Reagan administration actions to reduce NHTSA's budget and ease safety regulation of cars and trucks.

They see the report as providing ammunition for their side as the administration proposes to reduce funding for NHTSA safety grants to states from $189 million in the past fiscal year to $68 million in the current one.

The report, "Traffic Safety Trends and Forecasts," projects an increase in the rate of highway fatal-ities based on several assumptions:

* A growing concentration of small cars and light trucks combined with an increase in the numbers of heavy trucks is expected to increase motor vehicle fatalities by 10,000 a year by 1990, according to the study, because of the greater risk of fatalities in collisions between very large and very small vehicles.

* A greater percentage of the eligible population holding driver's licenses, combined with expected population growth, is expected to add 5,000 highway deaths a year by 1990.

* The mileage driven by each licensed driver is expected to increase from an average of 10,400 miles per year in 1980 to 11,100 in 1990, adding 4,000 fatalities annually by 1990.

* An increase in the use of motorcycles, mopeds and bicycles, now growing at a faster rate than the population generally, will add 2,500 deaths a year by 1990, the study estimated. Motorcycle fatalities have risen by 61 percent between 1975 and 1980, when the number of motorcycles in use grew by only 17 percent in the same period, the study said, blaming the escalation on the repeal of mandatory helmet-use laws in many states.

* Pedestrian fatalities, which reached a peak of 10,200 in 1972, declined to 8,096 in 1979, a drop attributed by the study to a decline in driving due to higher fuel prices. The total is expected to rise slightly by 1990.

* Fatalities could rise by another 5,000 a year at the end of the decade depending on whether states repeal the current 55-miles-per-hour speed limit on major highways, as some states have proposed.

Partially offsetting this trend is the change in the age mix of drivers, leaving a smaller percentage of motorists under 25 years of age. This could reduce fatalities by 3,000 annually by 1990, the report said.

In all, the NHTSA study projected that the total of 51,000-plus motor vehicle fatalities in 1980 could rise to 70,000 at the end of the decade.

The authors of the study also make the critical assumption that usage of safety belts will not increase significantly. Now, fewer than one motorist in nine wears safety belts, according to other NHTSA studies, nullifying the most important safety feature in automobiles. With 100 percent usage, 12,000 lives could be saved annually, the study concluded.

Peck is preparing an big public relations campaign to persuade motorists to wear belts. At the same time, he is considering dropping or modifying a major NHTSA regulation requiring installation of "passive" safety belts in future new-car models to restrain a motorist automatically as the car door closes. Opposing sides have debated whether a publicity campaign would be as effective as a strong, mandatory automatic-belt regulation.