I don't know about the EPA test but the European emissions test is a three minute simulated drive on a rolling road FROM COLD.
This therefore favours small petrol engines that warm up fast and also favours hybrids and EV's as the total distance travelled is less than 10 miles.
In the real world it is often the case that larger diesel engines will do better than their 'official' figures because they have warmed up to normal operating temperature and have reached optimum efficiency. Also a big torquey diesel engine suffers less dramatic loss of mpg when hauling a loaded vehicle, extra passengers etc. So basically the European emissions test is next to useless for giving you a good feel for what the vehicle would actually get in real world driving.
In my case, my 2.5TDi B2500 pickup truck has consistently bettered the official figure and by quite some margin in many fillups whereas all the small engined vehicles I've ever driven have never reached their official figures.
Diesel engines now make up over 50% of vehicle sales and in the UK at least the company car fleet users tend to opt for bigger 2.0ltr engines which give excellent mpg at motorway speeds where many of these fleet cars spend their life.
I have been getting a lot of 77mpg tanks recently and I have a fairly small diesel engine (1.4 litres, 10 years old). It does not have a DPF due to it's age. I think the DPF messes up a lot of people's MPG figures because the NEDC does not include a regeneration.
Don't believe estimates, necessarily. My 2005 Neon with its 2 litre engine averages 26 mpg with automatic trans. It was rated at 25/32. I think I "should" be averaging 28 to 30 mpg based on where and how I drive but I don't. If you really want good mileage drive a small car in winter and ride a small displacement motorcycle the rest of the time. For the record, I have other vehicles that meet their mileage estimates but they're all motorcycles. One was right on target, but that was one vehicle in 35 years and it was a bike. I have learned over the years that government mileage estimates tend to exaggerate car mileage and bike mileage is more accurate. Why? Politics?
I have another comment. It seems to me many of you are exaggerating your mileage. I don't believe for a minute someone can average more than 50 miles per American gallon with a car whether it's a diesel or not, and I'm talking about real cars, not 50 HP puddle jumpers with 12-inch wheels or hybrids. What makes me think this? The Honda CRX of a generation ago had a tiny engine, fuel injection, computerized ignition, and couldn't get 50 mpg, but it's a gas sipper to this day if you can find one. If you can beat a CRX mileage-wise with a conventional automobile I will humbly eat crow. As far as I can see your best bet mileage and cost wise for a car is a Toyota Corrolla with manual gearbox and 42mpg. Can't beat it. Best buy, and I'm not a Toyota fan because they rape you at the garage. Domestic cars are best for that, even if they get less mileage. That is why my 1986 Sable wagon cost the same as the 1985 Honda Civic hatchback I had before to own and operate over consecutive 6-year periods. The fact I owned the Sable after put it at a disadvantage but the overall cost was the same even though I was averaging 36-38mpg with the Honda but 22mpg with the Sable, and the Sable was a much nicer, roomier car the whole time. Chew on that and get back to me.
I had a manual Neon rated 28/35 and, after 30,000 miles or so (break-in maybe?), I started getting low to mid 40mpg for my mostly highway commute. If I used the air conditioner my mileage would drop to high 20s/low 30s so I never used it. I bet I could have gotten 50 in the CRX.
My motorcycle also gets better mileage than it is supposed to. I was getting 60s but now am only getting in the 50s.