It's a 2000 Accord 2.3L
210,000 miles.

Every 3 months I have to add about 1/2 gallon of antifreeze or it will overheat from low levels when the car idles.

I pressure tested the system and I could not find any leaks or air noise. Pumped to 16 psi it loses 1 pound about every 20 seconds.

I just changed the oil and did not see any milkshake type oil, just black. I would think 1/2 gallon of antifreeze in the oil pan would be very noticeable??
I don't know what to do next. should I just try again and look harder for leaks? I need your help pleeeeeease.

 

So no pinhole leaks when the system is pressurized? If not then the leak is likely internal, into the combustion chamber, and the coolant is getting burned. The symptoms of a leaking headgasket aren't always oil-in-water or vice versa, especially with a small leak. Sometimes the car misses for a few seconds after a cold (overnight) start on one or more cylinders due to coolant fouling the plug(s). A leakdown test will tell if one or more of the cylinders is leaking coolant past the headgasket. When the affected cylinder(s) is pressurized @ TDC, coolant will be pushed back through the radiator.

 

That's exactly what my car does on cold startups. didn't think of the two being related. Ok so I'll do a leakdown test next and go from there. Thanks for your help, makes sense now.

 

when it's COLD, remove the radiator cap and start the engine. if you see bubbles or agitation in the coolant then you're the proud owner of a bad head gasket. i'm almost positive that's what it is anyway.

 

I agree, it's likely a bad head gasket, but I never saw any bubbles on mine, just a mysterious coolant loss. Even during the leakdown test there was no air pushing into the radiator; just coolant getting pushed out of the radiator filler. IDK why that is. While the engine was running coolant got sucked into the cylinder but combustion gasses never got pushed back into the cooling system. Like a one-way valve. Weird.

 

My diagnostic approach: get, or borrow a compression test unit (AutoZone?). take out all the plugs,
spin motor, move unit to next cylinder, repeat and write down compression in each cylinder.

 

When you do the leakdown test, air leaking into the cooling jacket will make an air pocket right there in the cooling jacket. The engine isn't running, so that air won't be circulated around to the radiator. All you see is the coolant rising in the radiator neck.

Coolant getting sucked down when you start the engine is a bit confusing...?? I've seen that when the headgasket ISN'T leaking, but maybe the headgasket leak is small-enough that it still happens?

 

That makes perfect sense. I guess if I would have left the cylinder pressurized for a long time the air would have eventually made its way into the radiator.

I worded that badly. I'm guessing what's happening is the cooling system is hot & pressurized, the engine is shut off, and over several hours coolant is pushed past the bad head gasket into the combustion chamber.

 

No you didn't word it badly, but here's what I was thinking...

Even without a leaky headgasket, you can open the radiator cap & start the engine. The coolant level in the radiator neck drops down; I've seen that but don't quite understand why it sucks down.

But if the headgasket leaks, why doesn't the coolant rise in the radiator neck when you first start the (cold) engine?

I suppose it still sounds like a headgasket, even though it doesn't show EVERY possible symptom.

 

I'm guessing the initial spin up of the water pump drops the coolant level temporarily. I'm baffled too: the leaky head gasket I had seemed to leak one way only, from the coolant passages to the combustion chamber, never the other way.