Here's an important thought to consider when figuring out how to maximize your economy: some vehicles are most efficient at speeds above the speed limit. This is a highly counterintuitive concept but it does make sense, and here's why. Your engine is most efficient within a certain range of RPMs. Usually this is somewhere in the middle of the RPM range, often a bit above that. While at very high RPMs you do lose quite a bit of efficiency due to friction, this is not such an issue at nominal running speed. Meanwhile, your transmission trades speed (RPMs) for torque in all gears except your overdrive gear, in which it trades torque for speed (revolutions per minute, measured at the tires.) Thus, provided your car is appropriately aerodynamic, your best efficiency in miles per gallon (barring the squirt-throttle technique) can often be found in the middle of your engine's powerband, and in overdrive.
In my 1989 Nissan 240SX this is at about 80 miles per hour in fifth gear - this is where I get my best freeway mileage (approximately 30 mpg.) In my 1981 Mercedes 300SD Turbo-Diesel, it's also at about 80 MPH. Note that both of these cars are geared fairly low, because neither of them has much power compared to other vehicles in their class; this means that they are closer to their maximum engine RPM rating at cruising speed than many other cars. Both vehicles were also designed specifically for aerodynamics; the 240SX Fastback in particular has a Cd around .26.
Do not consider this an incitement to speed, only food for thought.
By the way, engine braking doesn't necessarily help your fuel consumption any. To understand why this is, you need to understand how the engine decides to deliver fuel. On any vehicle with an oxygen sensor (aka O2 sensor) the driver decides how much air to admit to the engine (if that - more on this later) and the computer decides how much fuel to deliver. On carbureted (yuck) vehicles, this is accomplished by the mixing control, which could be a solenoid but is usually a motor that continually opens and closes a port. On fuel-injected vehicles, this is controlled by changing the fuel pressure, pulse width, or just the duty cycle of the fuel injectors.
Either way, the PCM (powertrain control module - used to be called ECU, or engine control unit) is constantly monitoring the oxygen sensor. The O2 sensor, when heated, generates electricity because oxygen ions are attracted to it, and pass across it, inducing a current. More voltage means a leaner mixture, which is to say that it has more oxygen in it because less of the oxygen has been burned. Less voltage means a richer mixture, which means more of the oxygen has been burned. When the PCM detects that the mixture is lean, it adds fuel; when it is rich, it reduces it. This is going on continually while you drive, once the vehicle heats up (generally, as detected by the coolant temperature sensor.) If you hook the O2 sensor output up to a digital oscilloscope you can watch it draw a neat litle wavy line on your screen, rising and falling up to a few times a second.
Whether you are in overdrive or in second gear, when you let up on the accelerator pedal, the throttle butterfly valve (a round valve which closes off the air intake) is closed. This is where I clarify the point raised earlier about the user determining how much air enters the engine; in modern vehicles you may not even get to do that much. On vehicles which use GM's NorthStar engines, they use throttle-by-wire, which means that the only thing connected to the pedal is a potentiometer (the same kind of thing used for a volume knob) and the butterfly is controlled by a servo, as is fuel delivery. Even on some cars without throttle-by-wire, there is a secondary butterfly, usually located at the intake ports on the cylinder heads, which the PCM can use to decrease the air intake.
If you are engine braking, then the engine has to run faster than it does when you are not engine braking. Except in a few vehicles, this does not result in fuel and spark being shut off. Generally speaking, each revolution with the throttle closed uses a fairly finite quantity of fuel. As a result, engine braking will cause most vehicles to consume more fuel. This doesn't make engine braking useless; it saves your brakes. Most importantly, if you're not using your brakes they don't heat up, so engine braking continues to be an absolute necessity when descending long, steep hills. And, ultimately, it isn't all that much more. But the point is that the PCM is trying to run as close to a stoichiometric burn ratio as possible (about 14.7:1 ratio of air to fuel by mass at sea level with nominal temperature and barometric pressure) and so with the throttle plate closed, that's going to be more fuel per second when the engine turns at a higher number of RPMs.
Some vehicles DO cut fuel during engine braking; this is apparently common in modern BMWs. If you think your car does it, here's a quick test that you can do in vehicles with a manual transmission: get yourself on a nice straight, flat piece of road with no one else around, and get up to speed. Now, let out the gas pedal - you are now engine braking. Now, turn off the ignition. This will stop your power steering, power brakes, et cetera, so be sure to do this in a safe location, and don't blame me if you get killed! If the car's engine braking slows you down more with the ignition off than with it on, that probably means that your car is not doing a fuel cut. This is probably not a guaranteed test, because there's any number of tricks that can be done to change the way the car behaves, and on a modern vehicle with an automatic transmission you actually have no control over the transmission itself. Instead of using a manually-actuated valve body the shifter is attached only to a switch and the valves are actuated by solenoids controlled by a small computer, which is usually separate from but still talks to the PCM.
Now a couple quick notes on other things you can do to improve mileage: There are a number of additives which claim to do this, including acetone which can be purchased at any hardware store. Don't run more than 5% acetone, and don't come to me if it does something bad to your car, although the worst thing it should do is maybe damage your seals. That's bad enough, so if you're not sure, don't do it. Also, while most aerodynamics modifications are targeted at increasing downforce and thus drag, not all of them fit into this category. In particular, kits for third and fourth generation corvettes sometimes include an underpan which reduces drag on the underside by reducing turbulence, which is done in order to reduce lift (corvettes of these generations produce a great deal of front-end lift at high speeds, which makes the car very dangerous to drive fast without modification.) And along the lines of the note on removing your mirrors above, there are "aero mirrors" which are simply smaller mirrors with which you can replace your factory mirrors. They are available for basically all sports cars, and anything that is commonly modified (like the Honda Accord.) The mirrors are often an incredibly high portion of the drag of these small, aerodynamic vehicles and replacing them can make a substantial difference in Cd.
On the issue of air conditioning versus open windows: It should be obvious but is seldom stated that if you are going under a certain speed it is more efficient to have your windows open, because drag plays a bigger part in efficiency at higher speeds. It should be equally obvious that open windows represent different amounts of drag not just at different speeds, but also in different vehicles and even in different wind conditions. Driving around town, it may well be more efficient to have your windows down. Get on the freeway, and it's time to roll them up... unless you're only doing 55 in some massive barge of a truck, in which case it might not make any difference at all.
Finally, a note on the subject of drag in pickups: most pickup trucks get better mileage with the tailgate up and the bed empty than with no tailgate, the tailgate lowered, the tailgate replaced by a net, or even the bed covered (e.g. with a tonneau cover.) The theory is that air passing over the cab and falling behind the truck creates a pocket of turbulence in the bed that functions like an airfoil, allowing the wind of the vehicle's passage to flow smoothly over the rear of the vehicle. It would be a mistake to modify a pickup truck's shape to reduce drag without performing testing in your specific driving conditions.