Mobil commercial claims its oil "has been in more Indy 500 winners than any other oil." Quaker State shows an engine with a terminally corroded inside what they imply could happen when you use another oil. Exxon's commercial for its Superflo oil urges motorists to "rely on the tiger."

Oil companies spend millions of advertising dollars each year to convince you that their oil can make your car's engine perform better and last longer. And purveyors of motor-oil and engine "treatments" assert that their products offer engine protection that oil alone can't provide. In our most ambitious test project ever, we set out to discover whether such claims are fact or fancy.

One way to gauge the performance of motor oils is to test them on the road. We did just that, using a fleet of 75 New York City taxicabs. Indeed, the oil industry itself tests its oils in New York City taxis.

For 22 months, we tested the performance of 20 popular motor oils. Each of those oils met the industry's latest standards, as certified by a starburst symbol on the container.

We also tested Slick 50 Engine Treatment and STP Engine and Oil Treatments.

In addition to the taxicab tests, we had the oils' chemical and physical properties analyzed by an independent lab. We also surveyed our subscribers about their oil-changing experiences and preferences, and we sent shoppers to quick-lube centers across the country to assess the service

We put identical rebuilt engines with precisely measured parts into the cabs at the beginning of the test, and we changed their oil every 6,000 miles. That's about twice as long as the automakers recommend for the severe service that taxicabs see, but we chose that interval to accelerate the test results and provide worst-case conditions. After 60,000 miles, we disassembled each engine and checked for wear and harmful deposits.

Our test conditions were grueling, to say the least. The typical Big Apple cab is driven day and night, in traffic that is legendary for its perversity, by cabbies who are just as legendary for their driving abandon.

When the cabs aren't on the go, they're typically standing at curbside with the engine idling - far tougher on motor oil than highway driving. What's more, the cabs accumulate lots of miles very quickly, making them ideal for our purposes. Big-city cabs don't see many cold start-ups or long periods of high speed driving in extreme heat. But our test results relate to the most common type of severe service - stop-and-go city driving.

Each of the 20 oils we studied was tested in three cabs to provide meaningful test results even if a few cabs fell out with mechanical problems or because of accidents. (Six of the 75 engines did, in fact, have problems, none apparently related to the oil's performance.)

Our shoppers all across the country bought hundreds of quart containers of oil. Some brands had slightly different formulations in different areas, but all the oils included a full package of additives.

The independent lab helped us identify the most representative formulations of each brand. Our engineers transferred containers of that oil to coded 55-gallon drums and hauled them to the fleet garage for testing.

Ideally, oil should be thin enough to flow easily when the engine is cold and remain thick enough to protect the engine when it's hot. The lab analyses of each oil's viscosity characteristics - its ability to flow-indicate that motor oils have improved since 1987, when we last tested them. This time, far fewer test samples failed to meet the viscosity standards for their grade - and those were typically outside the limits by only a slight amount. No brand stood out as having a significant problem.

We tested oils of the two most commonly recommended viscosity grades - 10W-30 and 5W-30. Automakers specify grades according to the temperature range expected over the oil-change period. The lower the number, the thinner the oil and the more easily it flows.

In 5W-30 oil, for example, the two numbers mean it's a "multiviscosity" or "multigrade" oil that's effective over a range of temperatures. The first number, 5, is an index that refers to how the oil flows at low temperatures. The second number, 30, refers to how it flows at high temperatures. The W designation means the oil can be used in winter.

A popular belief is that 5W-30 oils, despite their designation, are too thin to protect vital engine parts when they get hot. However, one of our laboratory tests measured the viscosity of oils under high-temperature, high-stress conditions and found essentially no difference between 5W-30 oils and their 10W-30 brand mates. But at low temperatures, the 5W-30 oil flowed more easily.

Viscosity grade is important, so be careful. Recommendations vary with the make, engine, and model year of the car, so check your owner's manual and ask the mechanic for the proper grade of oil.

Of the 20 oils we tested, nine were conventional 10W-30 oils, and eight were 5W-30. We also tested two synthetic oils, Mobil 1 and Pennzoil Performax, and one synthetic-and conventional blend, Valvoline DuraBlend; all three were 10W-30 oils.

If you've been loyal to one brand, you may be surprised to learn that every oil we tested was good at doing what motor oil is supposed to do. More extensive tests, under other driving conditions, might have revealed minor differences. But thorough statistical analysis of our data showed no brand-not even the expensive synthetics-to be meaningfully better or worse in our tests.

After each engine ran about 60,000 miles (and through 10 months of seasonal changes), we disassembled it and measured the wear on the camshaft, valve lifters, and connecting-rod bearings. We used a tool precise to within 0.00001 inch to measure wear on the key surfaces of the camshaft, and a tool precise to within 0.0001 inch on the valve lifters. The combined wear for both parts averaged only 0.0026 inch, about the thickness of this magazine page. Generally, we noted as much variation between engines using the same oil as between those using different oils. Even the engines with the most wear didn't reach a level where we could detect operational problems.

We measured wear on connecting rod bearings by weighing them to the nearest 0.0001 gram. Wear on the key surface of each bearing averaged 0.240 gram - about the weight of seven staples. Again, all the tested oils provided adequate protection.

Our engineers also used industry methods to evaluate sludge and varnish deposits in the engine. Sludge is a mucky sediment that can prevent oil from circulating freely and make the engine run hotter. Varnish is a hard deposit that would remain on engine parts if you wiped off the sludge. It can make moving parts stick.

All the oils proved excellent at preventing sludge. At least part of the reason may be that sludge is more apt to form during cold startups and short trips, and the cabs were rarely out of service long enough for their engine to get cold. Even so, the accumulations in our engines were so light that we wouldn't expect sludge to be a problem with any of these oils under most conditions.

Variations in the buildup of varnish may have been due to differences in operating temperature and not to the oils. Some varnish deposits were heavy enough to lead to problems eventually, but no brand consistently produced more varnish than any other.

The bottom line. In our tests, brand didn't matter much as long as the oil carried the industry's starburst symbol

Beware of oils without the starburst; they may lack the full complement of additives needed to keep modem engines running reliably.

One distinction: According to the laboratory tests, Mobil 1 and Pennzoil Performax synthetics flow exceptionally easily at low temperatures - a condition our taxi tests didn't simulate effectively. They also had the highest viscosity under high-temperature, high-stress conditions, when a thick oil protects the engine. Thus, these oils may be a good choice for hard driving in extreme temperatures Note, too, that a few automakers recommend specific brands of motor oil in the owner's manual. You may need to follow those recommendations to keep a new car in warranty.

The long-time mantra of auto mechanics has been to change your oil every 3000 miles. Most automakers recommend an oil change every 7,500 miles (and a specific time interval) for "normal" driving, and every 3,000 miles for "severe" driving - frequent trips of less than four or five miles, stop-and-go traffic, extended idling, towing a trailer, or dusty or extremely cold conditions. Many motorists' driving falls into one or more of those "severe" categories.

In our survey, almost two-thirds of our readers said they had their oil changed every 3,000 miles or less. They may be following the thinking expressed by one of our staffers: "I have my oil changed every 3,000 miles because that's what my father did, and all his cars lasted for many years."

To determine whether frequent oil changes really help, we changed the oil in three cabs every 3,000 miles, using Pennzoil 10W-30. After 60,000 miles, we compared those engines with the engines from our base tests of the same oil, changed every 6,000 miles. We saw no meaningful differences. When Mobil 1 synthetic oil came out, Mobil presented it as an oil that, while expensive, could go 25,000 miles between changes. That claim is no longer being made. But Mobil 1 is still on the market, selling at a premium (along with pricey synthetic competitors from several other companies). And synthetic oil's residual reputation as a long-lasting product may still prompt some people to stretch their oil changes longer than the automaker recommends.

Determining whether synthetic oils last longer than conventional ones would require a separate test protect. To try to get some indication, we put Mobil 1 synthetic into three cabs and changed their oil every 12,000 miles.

We intended to compare the results of these tests with those from the three taxicabs whose Mobil 1 was changed at our normal interval, every 6,000 miles. Unfortunately, two of the three engines using the 12,000-mile interval developed problems. (We couldn't attribute those problems to the oil.) The third engine fared no worse than the three whose oil had been changed at 6,000-mile intervals.

The bottom line. Modern motor oils needn't be changed as often as oils did years ago. More frequent oil changes won't hurt your car, but you could be spending money unnecessarily and adding to the nation's energy and oil-disposal problems.

Even in the severe driving conditions that a New York City taxi endures, we noted no benefit from changing the oil every 3,000 miles rather than every 6,000. If your driving falls into the "normal" service category, changing the oil every 7,500 miles (or at the automaker's suggested intervals) should certainly provide adequate protection. (We recommend changing the oil filter with each oil change.)

We also tested Slick 50 and STP Engine Treatments and STP Oil Treatment, each in three cabs. (Slick 50 costs $17.79 per container; STP Engine Treatment has been discontinued.) All three boast that they reduce engine friction and wear.

The engine treatments are added with the oil (we used Pennzoil 10W-30). They claim they bond to engine parts and provide protection for 25,000 miles or more. We used each according to instructions.

The STP Oil Treatment is supposed to be added with each oil change. It comes in one formulation (black bottle, $4.32) for cars with up to 36,000 miles, another (blue bottle, $3.17) for cars that have more than 36,000 miles or are more than four years old. We used the first version for the first 36,000 miles, the second for the rest of the test-again, with Pennzoil 10W-30.

When we disassembled the engines and checked for wear and deposits, we found no discernible benefits from any of these products.

The bottom line. We see little reason why anyone using one of today's high-quality motor oils would need these engine/oil treatments. One notable effect of STP Oil Treatment was an increase in oil viscosity; it made our 10W-30 oil act more like a 15W-40, a grade not often recommended. In very cold weather, that might pose a risk of engine damage.

On the basis of our test results, we think that the commonly recommended 3,000-mile oil-change interval is conservative. For "normal" service, 7,500-mile intervals (or the recommendation in your owner's manual) should be fine. Change the oil at least that often to protect your engine and maintain your warranty. Even for the severe service experienced by the taxis in our tests a 6,000- mile interval was adequate. But some severe service - frequent cold starts and short trips, dusty conditions, trailer towing - may require a shorter interval. Note, too, that special engines such as diesels and turbos, which we didn't test, may need more frequent oil changes.