Ellis Parsons looked out over New York Avenue and Bladensburg Road NE, the growl of afternoon traffic swelling around him, and said, "You couldn't pay me $100 to stand near that intersection during rush hour."

Parsons speaks from three years' experience running a gas station at the Washington area's most accident-prone intersection -- a maze of 24 traffic lights, eight lanes of east-west traffic, four lanes of north-south traffic, two pairs of left turn lanes, concrete median strips, metal accident barriers and deflectors filled with sand and water.

Among the 90 accidents reported there last year were two cars and a truck, coming in for landings west-bound on New York Avenue, that whistled across his Texaco station lot, leaving a smashed car, an uprooted fire hydrant and startled employes in their wake. So many cars have made unscheduled stops against fenders and doors of autos at his station that Parson's employes now make certain their own cars are parked in back.

The curbside traffic light out front has been bowled over so many times Parsons has lost count, although he's particularly fond of the time city employes spent several hours last year replacing the pole-mounted signal, only to have it knocked down by a wayward motorist the next day.

"When I bought this station, I got nervous every time I heard a crash," said Parsons, cigar clenched between his teeth and a smudge of motor oil on his green uniform.

But now "if I hear squealing tires and a loud bump, I might look up or I might not," he said. "I guess I've just come to the conclusion that there will always be a lot of accidents at this cormer as long as there's so much traffic."

Traffic engineers agree and say that is the case with most bad intersections. The massive volumes of daily traffic have grown beyond the limits of modern safety techniques, they say, and point as evidence to the rogues' gallery of intersections making up the rest of the Washington area's "worst 10".

Those intersections, based on reports by police and highway departments, are, in order of total accidents:

Alexandria's Seminary Road and I-395, holding down second place; 3) Branch Avenue and Allentown Road, Camp ysprings; 4) Stanton Road and Suitland Parkway SE; 5) Beauregard and King streets, Alexandria; 6) Pennsylvania and Branch avenues SE; 7) Landover and Brightseat roads, Landover; 8) I-295 and Portland Street SE; 9) Maryland Rte. 355 and Shady Grove Road, Gaithersburg; 10) South Dakota and New York avenues NE.

"We have done everything we could over the last 10 years to improve traffic safety at New York and Bladensburg, but nothing seems to work," particularly with an average traffic volume of about 85,000 vehicles a day, said Alesandres L. Perkins, deputy assistant director of the D.C. Bureau of Traffic Engineering.

Since 1972, the D.C. Department of Transportation has spent more than $50,000 there to install overhead signs, additional turn signals, better lighting and water- and sand-filld devices to deflect cars hit in an accident. In addition, traffic light lens sizes were increased from eight inches to 12 inches.

"Right now we are at an impasse," Perkins said. "The only thing left for us to do is to take one street over of under the other. And that would be too costly."

Safety engineers elsewhere in the area have the same kind of problems. At Seminary Road and I-395, for example, officials had hoped an extended median installed two years ago would cut down on accidents as cars jockey for position exiting the freeway and entering Seminary Road. But that proved to be little help and now they say the only remedy would be the unthinkable -- traffic lights at an interchange.

At Beauregard and King streets, $400,000 has been spent over the last few years to widen them and improve synchronization of signals, but so far the effect appears to have been minimal, according to city officials, although they say it may be a little too early to assess the impact of some fairly recent changes.

And out in Maryland at Branch Avenue and Allentown Road, the state spent $55,000 three years ago to install longer turn lanes and improve the signal system for turns against heavy oncoming traffic, but with no major effect, according to a state official.

"We're seeing an alarming increase in intersection collisons nationwide," said Neal Boot, a spokesman for the National Safety Council. "I think the problem is a combination of a lot of things like large numbers of unqualified drivers and roads designed back when cars were bigger and the volume of traffic was much less."

Boot also said that the sight lines of road signs and traffic signals -- designed for the big cars of the 1950s -- have remained the same while automobiles have grown steadily smaller.

"How is the guy in a Honda supposed to see the traffic signals if he's stuck between a Cadillac and an Oldsmobile?" Boot asked.

For the most part, Boot added, traffic safety is an art that offers no guarantee that costly intersection improvements will achieve the desired results.

"All an engineer can do is go to an intersection and diagram what he believes to be the problem," Boot said. "Then he recommends improvements based on his theory of what is causing the problem -- like a doctor prescribing a medication. But there are no guarantees that the improvements will solve the problem. You might even opne a new can of worms by making some improvements. You might stop right angle crashes and start a series of rear-end collisions."

Traffic engineers keep on trying, however, because the stakes can be enormous.

Nationally in 1980, automobile crashes -- most of which occurred at traffic intersections -- cost the American public more than $20 billion in medical aid, rehabilitation services, lost wages and other costs, according to the Insurance Institute for Highway Safety.

In the case of New York Avenue and Bladensburg Road, accidents last year cost roughly $374,000, based on National Safety Council figures of $4,700 for an injury and $670 for a non-injury accident. Officials say it's impossible to calculate the cost of the host of unreported accidents.

There were no fatalities at the intersection, and, in fact, of the area's 10 most accident-prone the only intersection to have a fatality was Portland Street and I-295, where there was one.

That kind of accident statistic is fairly typical, according to Neil Pederson, an transportation engineering consultant for J.H.K. and Associates in Alexandria, who siad the bulk of the severe traffic accidents do not occur at intersections, but on heavily traveled freeways and high-speed bridges.

That is borne out in the Washington area, where some of the most dangerous stretches of road are parts of the beltway -- the Woodrow Wilson and Cabin John bridges and the section between Georgia and Wisconsin avenues.

Traffic engineers' efforts to reduce intersection crashes generally begin, as they do in Maryland for example, with a study of locations that have a history of a certain number of accidents each year, according to Paul Jaworski, chief of the bureau of accident studies in the Maryland State Highway Administration.

In Montgomery County, and intersection must have at least 12 accidents before it is considered "hazardous." In Prince George's County, the magic number is 16. In D.C. the number is also 16.

"Once a hazardous intersection has been identified, a traffic engineer will recommend procedures that he believes will relieve either the traffic volume or the conditions that have caused many of the accidents," Jaworski said. "But we know from the beginning we cannot relieve all of the accidents because a certain percentage is due to mere human error."

Once the engineer has submitted his assessment of the problem, a proposed solution and the estimated cost, the state then compares the cost of the repair to the cost of losses to citizens from personal injuries and property damage.

If the proposed improvements would not be cost beneficial, law enforcement agencies are urged to increase their patrols of hazardous intersections, he said.

Despite the frustrations, traffic engineers can point to their share of successes, a case in point being Stanton Road and Suitland Parkway.

In 1978, that location, although it did not have the most accidnets, ranked first on the D.C. traffic engineers' list of hazardous intersections, under a complicated formula that places heavy emphasis on the severity of the accidents. Then the city spent $244,260 to install left turn lanes, improved traffic signals and high intensity street lights. The improvements paid off through a 44 percent reduction in accidents and a 76 percnet reduction in personal injuries, according to Perkins.

At the same time, D.C.'s experience at the intersection illustrates the limits and the heartaches of the traffic engineer's art. Thee had not been a fatality in the year before the improvements, but there was one the in the subsequent year; there has not been one since. And the number of accidents has started to climb again at the intersection, according to police statistics.

It is 5:30 p.m. and the rush hour has reached its peak at New Ork and Bladensburg. A tow truck with fancy lettering spelling out "Billy's Towing" sits like a giant condor, quietly observing the traffic as it squeezes through the intersection and waiting for its moment.

"I sometimes have my trucks just wait there," says owner Paul R. Tibbs Jr. "I know we'll be called sooner or later."