Stop - Engine Temperature warning

2.5 volts is not right,
all the sensors that run 5V SHOULD have 5V.
Disconnect all the 5V sensors and check 5V feeds to each,
Now monitoring the 2.5V one,
which should now read 5V,
re connect the sensors one at a time until the 2.5 returns.
whatever sensor it was that dropped the voltage is suspect.
The ECU produces the 5V reference internally but often this is shared by more than one sensor.
Something is eating your missing 2.5V

I tried as suggested and received a funny result. The "top" connector on the engine (3-pin blue connector, one green wire, one yellow/green earth wire) causes the strange behaviour for the other sensor.

As soon as I unplug this sensor, I get 5 Volts for both sensors. I measured both sensors and the top one has 8.4kOhm while the side sensor has 4.1kOhm (at about 14°C room temperature, engine is cold). I tried to put a 4k7 resistor into the top connector, this also causes the same malfunction (voltage dropping for the other sensor).

I also tried to follow the cables but couldn't find the other end. If I unplug the middle plug on the ECU, the error is gone. But there is probably a lot going on there ;-)

So, my question: How likely is an ECU failure? I found this now:
http://www.cartechelectronics.com/engine-ecu-bosch-me-7-4-4-or-m-7-4-4-p0118-coolant-temperature-malfunction-repairs/
They charge €182 for this problem, and this is exactly my ECU:
"Engine ECU Bosch ME 7.4.4 or M 7.4.4 P0118 Coolant temperature Malfunction Repairs"
(I'm from Austria, not UK, so I may use another supplier).

Sorry, my last posting was meant to be a "further question".

Alas colour codes for wire do not help Citroen and Peugeot use alpha numeric,
My wiring diagram shows the CTS to be M1220 for the 5V and 1235 for the return signal.
So you may be on the right sensor.
5V feed is from ECU pin 37,
this is split to feed the CTS/intake temp sensor/MAP sensor/throttle position AND the power steering switch,
Which is something I suspect you've ignored till now.
Disconnect the power steering switch and retest for 5V.
Failing that it time to load test the wire harness.
I use a 6V motorcycle battery leads and an indicator bulb/holder.
Unhook the multi at the ECU and back probe T37 Bulb plugged into the sensor output and the return at the ECU plug to test battery ground,
bulb should glow bright,
make sure the bulb V is correct for 6V.
Or strip the harness back,
near as I can tell there's 4 splices on the 5V line from ECU T37 plus the power steering switch,
Any of which could be a high resistance connection.
Or If cash isn't a problem buy a set of "Load Pro" leads and learn to use them.
Definitive lead set for a multi meter,
Look up "LOAD PRO" on Youtube
This will test for high resistance which you do have,
there's no other explanation for 5V to become 2.5V

Dear 1/4drive,
Sorry for not getting back soner and thanks a lot for your comprehensive answer.
I found another "solution". I opened the ECU case and tried to pry out the ECU board using a screwdriver. It didn't go very far and quickly I realized that the connectors are rather well connected to the upper half of the ECU housing, as well as being soldered to the board. Therefore I just gave up and put the ECU back into my car. The problem was gone!! I can only imagine some bad contact inside my ECU. I had unplugged and replugged the ECU plugs at least three times without any effect, so I don't think it was any of the cables or plugs. There was also no sign of corrosion.
I've been using the car for two weeks now and everything is fine. Thanks for helping out!.