Oh, the irony.

It seems every time the EPA appears to have its act together in offering mileage numbers that you and I just might be able to achieve, a new calamity erupts with the car companies' self-reported results to cloud the public's faith in them all over again.

The most recent incident was the Ford C-Max's original 47-mpg city and highway claim, which came under fire almost instantly. The EPA's astonishingly loosey-goosey rules had allowed the Fusion Hybrid's results to be used for the C-Max. (A loophole permitted Ford to recycle data from its biggest-selling offering in the same weight class using the same drivetrain.)

No wonder, then, that so many new-car buyers remain wary of those big, bold numbers on window stickers. Are you among them? Great timing, because it just so happens that we're inaugurating a brand-new alternative source for mileage info: Motor Trend's ambitious real-world program we call Real MPG.

Doing all the hard and complicated work to create these is our British partner, Emissions Analytics, operating out of our vehicle studio in El Segundo, California, and headed up by the ever-cheery Sam Boyle and his assistant Jesus Flores. To explain how our Real MPG numbers are created, let's tag along with them on an actual test. And to spice things up, the car they're driving is none other than the controversial C-Max -- which, by the way, is a car we really like, no matter what mileage it produces. How will it do? We're using all the technical firepower we have to find out.

In Frank Markus' "Technologue" column, he explains the technical details of how EA's sophisticated electronics sniffs once-a-second exhaust samples for CO and CO2 and calculates the fuel's burn rate by multiplying it by the total exhaust flow. In practice, it takes about 90 minutes to instrument each car, from building up the plumbing that delivers the exhaust to the flow-rate analyzer (mounted on the car's rump, where it's tucked out of the airflow) to threading a tube carrying exhaust samples inside the cabin to a gas analyzer. A mini weather station monitors environmental factors, and a GPS data-logger notes speed and position. There's even a separate electrical power supply (a dang big battery) to isolate the system from imposing parasitic losses on the car. Moreover, before and after every test, the analyzer is checked: first, with a flow of nitrogen (as a "zero") and then with one containing precisely 16 percent of CO2 to check its calibration.

The fuel Sam pours into the car's tank before each test is as important as anything else. You'll be surprised to learn (we were) that the energy density of gasoline can vary by 4 percent or so between deliveries of what's outwardly the "same" fuel to the same gas station. That's an intolerable variable for mileage testing. Which is why the EPA, as well as such premier testing facilities as Argonne National Lab, use a "test" gasoline that costs up to $100 per gallon. Eeek. Instead, we've partnered with the good folks at Chevron who, at their Richmond refinery in the Bay Area, are brewing up a trio of standardized, real-world "typical" fuels (regular and premium grades with just under 10 percent ethanol and, of course, diesel) that'll remain consistent and chemically stable for years. Periodically, 55-gallon drums of these are shipped south to another partner in this endeavor, Dion & Sons, in Long Beach, California, who carefully decant it into 5-gallon cans for delivery to our offices (thanks, Dave). To ensure consistency, Chevron captures a gallon per barrel for random post-analysis, and Dion & Sons quarantines a quart from every barrel it receives for the same purpose. The idea of going to the gas station seems so quaint (and naive) now.

Sam and Jesus' route is a carefully crafted, 88-mile drive influenced by the EPA's city cycle. It travels along a nearby freeway, through myriad city streets, and on connector roads to illuminate the effect of elevation changes. And there's plenty of cleverness woven into its planning. The freeway portion (with a target average speed of 65 mph) is traversed six times (two with the A/C set on low, four with it completely off). In between, the 22-mph-average-speed, 22-turn city portion is repeated four times, twice each at those same extremes of climate settings. The repeats are for two reasons: to check for internal consistency and to quantify the impact of climate control on mileage, which can be quite substantial. Windows rolled up and no A/C? Sam and his team have cool vests at the ready to help them endure the occasional sweltering summer day. Our sacrifice for you, cherished reader.

Obviously, our Emissions Analytics pilots attempt to drive the cars in a highly consistent manner. Nevertheless, how do they account for variations in L.A.'s notorious traffic? Even at the quietest of times, we can't control that. But by modeling the data and seeing how the second-by-second CO2 varies with speed, acceleration, and gradient (and other) factors, small corrections can be made whenever the real driving deviates from our target test cycle. In most cases, the result is a high degree of comparability among different cars tested on different days, though if a profile differs too much, it means retesting. However, the golden lining here is that these smaller corrections give us valuable insight into how each vehicle reacts to changes in key driving behaviors (in particular, acceleration rate). It's amazing how much intelligence can be gleaned from a test that's not much over 3 hours, excluding setup time.

And what needs to be underlined here with a big fat yellow highlighter is that this is happening in the real world with actual aerodynamics, true rolling resistance, and real-traffic acceleration rates -- not on a dyno with simulated aerodynamics and clever test drivers following a wiggly line on a monitor that their right feet could reenact in their sleep. (The veracity of which was even questioned by the EPA's director of Transportation and Air Quality, who recently fretted that the dyno jockeys were getting a little too smooth at tracking the traces.)


How did our C-Max do?

After sifting and tuning the data -- for instance, its A/C load gets interpolated at 50 percent of full blast, data compromised by unusual traffic is deleted, the city and highway results are weighted per the EPA's 55/45 ratio, and there's compensation for the difference between the battery's pre- and post-test state of charge -- the C-Max got…45.6 mpg. For perspective, the average of 107 owners' self-reported results from Fueleconomy.gov is 39.2, while the average of 293 drivers at Fuelly.com is 40.7. Notably, our number still falls in the meat of the broad distribution of results.

The takeaway confirms what we knew going in: Hybrids are remarkably sensitive to driving habits, including those intrinsic to our Real MPG test cycle, but also the car's original EPA rating was clearly too high. When the grumbling about the C-Max's 47/47/47 EPA numbers began, Ford countered with "It's winter, the batteries are cold. You'll see its numbers climb as the weather warms." However, by August 2013, Dearborn had reluctantly done an about-face, downgrading the C-Max's numbers to 45 city and 40 highway (43 combined) and sending $550 to grumpy early buyers and $325 to lessees.

In this issue you'll find Real MPG results for Car of the Year finalists. And we're ramping up our testing pace, so they'll soon be populating our website, too. And there, in the magical world of interactivity, you'll be able to fine-tune a car's Real MPG result based on your specific driving behavior. Even for a car that's as super-sensitive to driving habits -- such as the Ford C-Max. Stay tuned.