I have a 94 Honda Civic DX. Last summer it became almost impossible to start in hot afternoon weather but the next morning (cooler) it would turn over without a problem. I did some research and found that it was most likely the Main Relay. I replaced it with a new Main Relay and the starting problem went away... for a while.

Seven months later the hard starting returned but it was now hard to start even in mornings and cooler weather. It also seemed to be most difficult after rainy or very damp weather. I would crank the car and the Check Engine Light would come on, stay on, and the car would not start (I did not hear the Fuel Pump power up after a couple of seconds like I had before). Now, here's where it gets strange, if I left the key in the ignition and turned to the second click, the Check Engine Light stays lite for about 2 to 3 minutes (not seconds) and then it goes off and THEN the Fuel Pump powers up (I hear it).

What is keeping the CEL on for those 2 to 3 minutes and delaying the application of power to the Fuel Pump until then. Its quite repeatable as long as the car is off for about 30 minutes between attempts. The Fuel Pump will power right back up if the car is turned off and then turned back on only a few minutes later.

Your ECM is failing. To confirm, perform the K-Test:

The K-Test: Remove the MAP Sensor connector and turn the ignition switch to ON (Position II). Using a multimeter, check for 5 volts going between the MAP Sensor connector's reference wire (+) and ground. As you look at the connector, this is the socket on the right. Really press the black test lead into a cleaned main ECM ground on the thermostat housing. If the voltage is low, it's probably indicating ECM failure. Most failed ECMs will record a fraction of a volt. To me, the K-Test is simple, elegant, and accurate.

Will the K-Test only show the ECM has failed (almost no voltage) during the interval the CEL is lite (that 2 to 3 minute waiting period until I hear the Fuel Pump power-up).

Once the unit powers up the car will start and must be powered off for some period (length I have not yet figured out) before the problem will occur again. Short intervals between powering off and another restart (5 to 10 minutes such as when running a set of quick errands) and the car will crank without any trouble. Its almost as if something must have sufficient time to either "cool down" or "discharge" before trying to start the car goes south again.

You likely have intermittent failure now, which proceeds total failure. The K-Test should show 5 volts all the time, in order for the ECM to be good.

Okay, I have been trying to measure the voltage but I wanted to do so while the car exhibits the problem. Problem is the problem comes and goes. For the last week (and I didn't drive the car all week, I only tried to test it, turning key to click 2) the fuel pump came up after only a couple of seconds and the CEL went off. All Week. Then a thunderstorm moved through (did not drive car) and the problem came back. 2 to 3 minute wait with the CEL on and then it went off and the fuel pump came up. I decided I would test the car, regardless, the next day (today). Okay, problem has vanished again, fuel pump comes right up and CEL goes off. I measured the voltage at the point specified anyway. Okay, my volt meter had readings bouncing all over from a few MV to several Volts, back and forth, never stabilizing. So, from your test that would say the ECM is dead... but today the fuel pump comes right up and the CEL goes off as if nothing is wrong even though the voltmeter is bouncing all over. To me, why is the pump performing normally if the ECM is dead, I'm confused. Not doubting you, just confused. Other point... I decided what the heck lets get a code out and see if that sheds any light on things. So I jumped the two pins to get codes and the CEL light comes on and stays on as long as the pins are shorted, no blinking just on. Remove the short and the CEL goes off. I also have two friends, both of whom know more about cars that I do, so I can't just dismiss what they say out of hand. They are both wondering if it might be a flaky/aging fuel pump. Is that a possibility or asked another way, is there some way to rule out this being a fuel pump issue? Help, please.

Jump the No. 5 and No. 7 pins to the main relay connector. If the fuel pump runs, when the ignition switch is turned to ON, it's likely fine.

I believe your ECM has not completely died, yet. However, it's moving in that direction. I would replace the ECM and be done with it. Look into buying a remanufactured ECM with a lifetime warranty.

Okay. I'm going with the assumption that the problem is a dying ECM because, from what you wrote it's the only thing that makes sense (the voltage reading is continuously cycling up from a few MV rising to about 2.5 volts, then back to a few MV rising back to about 2.5 volts, over and over).

Now, given that I am not very much of an auto mechanic at all, I have 2 questions about the ECM.

First: What will happen if the ECM reaches total failure? What I mean by this is will I no longer get the fuel pump to be able to power up on a start up attempt but that is the first I will be aware it has failed completely? Or if the ECM dies while the car is on the highway will the car go dead/quit right then and there (I will know the instant the ECM dies, rather than the first time I try to start it after the ECM dies)?

Second: On replacing the ECM. I have a general idea of where it is on my car (under the passenger side floor) but what is the best way to access it (so I can put everything back over it once it has been replaced) and how is something like this removed and replaced so I don't damage either connectors or the replacement ECM (I am assuming since this is essentially a computer it must be handled with extra care).

Like I mentioned in an earlier post on this thread, I do have 2 friends that are much more into automotive repair than I am but dealing with ECM problems goes beyond what either of them have done also so I am very much in the dark here.

Also, do you have a particular source you would recommend to get a replacement ECM. I went to O'Reilly's and, while they can get me one, they cannot promise how long it will take to show up. These kind of repairs must be something they rarely have to deal with.

1. CEL will stay on, fuel pump won't work, and the car won't start.

2. This should help you locate and replace the old ECM:

https://www.askmehelpdesk.com/cars-trucks/faq-how-troubleshoot-repair-maintain-hondas-selected-other-vehicles-46563-3.html#post235038

O'Reillys should be able to get you a replacement ECM within 1 day or less.

Thank you for your advice. I purchased another ECU and replaced the one on my car with that one. Have tested it a dozen times since (both before and after a thundershower) and the fuel pump comes right up as expected every time. My only problem has been with the passenger side floor molding strip that I had to remove. I can't get it to re-seat as tight as it was before I took it up. It doesn't want to stay flush against the metal under it. But I can live with that, the car is drivable again.

Nice job--I'm sure your are relieved.