Think about what is required to turn the engine over and prevent stalling.

-When the clutch is engaged whilst going down hill the wheels turn the engine. No Fuel Required. -When the clutch is disengaged the engine has to turn itself over - which needs fuel.

*Disclaimer, i'm sure carbourettas (sp?) could well be doing weird ass things so whilst conceptually the above holds they may well be different.

It's not the fuel required by the engine, it's the engine braking. Going down a hill with clutch disengaged, the Accord would accelerate. With clutch engaged, it would slightly decelerate. That's fuel that the engine doesn't need to burn to accelerate it back up to cruising speed.

Ditto when coasting to a stoplight - with clutch disengaged (or in neutral), it would coast for a longer distance, which - if you're careful and know your vehicle - means you can let up on the gas a lot sooner.

Anecdotal, but I did some tests in my previous car (a 2014 Mercedes, automatic) and it coasted really, really, really well. In drive, downhill, on the freeway, feet off all pedals, it would accelerate as much as if I put it in neutral, so I think its software and transmission allowed it to perform as good as a manual in that situation.

Not calling you out specifically, but a lot of people arguing for manual transmissions in this thread seem to base their experiences on ~10 year old cars or older, and weirdly assuming that nothing has happened since, that software and technology in cars has been magically frozen in time, despite enormous incentives for car manufacturers to improve it.

Automatics don't engine brake the same way, the torque converter is only fully engaged when it has some pressure buildup, so it will slip (and not maintain fuel cut) when coasting whenever the wheel speed won't maintain the pressure.

To test, watch your rpm when you left off the gas, if it drops to ~750-1000 rpm, your torque converter is slipping (as designed) and not maintaining fuel cut.

To agree with you, I think the comments in this thread started from a manual transmission standpoint and then everybody thought it applied to automatics.

It's not that simple, since everybody has had lockup torque converters (based on a computer-controlled clutch) for a long time now. As long as the TC is locked up, you have full engine braking. As you slow, it eventually drops out of lockup, you see the rpm drop to idle or close to it, and you feel the engine braking diminish (though you do still have a small amount, as the the transmission does have pumping and drag losses).

No free lunch.

In a manual you are burning fuel when the clutch is disengaged to keep the engine turning over. With the engine engaged you steal from momentum which lowers fuel use.

The mid point of keeping the engine engaged and adding enough fuel to reach your desired stopping point once the RPM enters the same range as an idling engine is actually most efficient.

I'm pretty far out of my element here, but as I understand even that isn't universal. BMW's Valvetronic for instance doesn't engine break the same way as other engines, because they don't have a throttle valve and don't have a vacuum outside the intake valves:

https://www.youtube.com/watch?v=DKfcV7hITYI

carburetor (American and Canadian spelling), carburator, carburettor, or carburetter (Commonwealth spelling)

https://en.wikipedia.org/wiki/Carburetor

True, but the amounts of fuel needed for idling is minuscule.

Basically any car that has electronic fuel injection does this. So at least the '99 Civic definitely does it.

Pretty much any engine that has an ECU I would suspect.

The oldest car I have that has an ECU is from 1987 (although judging the fuel efficiency on that car when it was new is a bit hard since the engine has been completely reworked since then -- including an aftermarket ECU, new injectors, new ignition etc which makes the car a tad more fuel efficient than my 2012 car)

My 1998 Saturn noticeably cut fuel on deceleration, it was easy to tell since the exhaust system was perpetually breaking

Come to think of it: the car I drove in 1988 cut fuel when coasting and it had a rather unrefined ECU mapping that would cause somewhat hard transitions from the fuel-cut state to feeding the engine go-juice.

I also have an 09 Fit. I use the clutch to coast all the time. I think the only gauge that shows fuel usage is the mpg one? when I coast it just goes over to infinite. I don't think it does that when I don't put the clutch in.

Had a manual 85 RX-7; it had fuel cutoff above 1000 rpm when coasting (drivetrain engaged)

92 was 25 years ago, this isn't a modern car, it should not be on the road if you care about economy and road safety.

Oh, it isn't on the road, my dad totaled it in 2008. It was, however, what I drove for my first 3 years in the workforce, so I got a pretty good feel for how to drive it economically.