I was going to write a long post in response to this, but I found a webpage, written by a Canadian CCP, that does a far better job:
http://www.paramedicine.com/pmc/End_Tidal_CO2.html
You've got to bear in mind with these situations that what we're really interested in is the amount of CO2 in the arterial blood, the PaCO2. CO2 diffuses very quickly, and if we consider an alveolus that's perfused, the partial pressure of CO2 in the alveolus (PACO2) is pretty much equal to the pressure in the surrounding capillaries.
* If we could always measure this value, it would be fantastic.
Unfortunately, when we're measuring capnography, we're also measuring the CO2 content of air from alveoli that are ventilated, but not perfused. These will have lower amounts of CO2 than are present in the arterial system, and will lower our reading.
[This discrepancy between perfusion and ventilation is often referred to as VQ mismatch]
As a result, in healthy people, our recorded ETCO2 is usually a little lower than PaCO2. Somewhere around 3-8 mmHg lower.
Now, this wouldn't be too bad. But a lot of the people we deal with are sick, especially the respiratory disease patients, and this difference can often be much greater, to a point where the PETCO2 reading is almost meaningless. We might have someone with a PaCO2 of 55, and a PETCO2 of 30. Our recording device is suggesting that we're hyperventilating the patient, but actually, they're hypercapnic. A simple situation may occur when cardiac output is decreased, e.g. cardiac arrest, shock, hemorrhage, etc. or perfusion is otherwise disturbed (e.g. PE).
We can also see the relationship go the other way, with a higher PETCO2 than the corresponding PaCO2 measured by ABG, if our patient is hypothermic -- as CO2 solubility differs with temperature, and ABGs are corrected to 37C.
There's a lot of research out there, I can link to some if you guys would like, although most of it requires a university log-in. But if you start looking at real numbers in patients seen in the ED or transported by EMS, there's a very poor correlation between ETCO2 and PCO2. So while these numbers are the best measure of ventilation that we have, I don't think they're as useful as we're often made to believe in paramedic school.