For the extraction of aminoacids from complex matrices, (i.e. coffee) intended for chromatographic detection, and having a hydrophobic C18 column, which solvents and which protocol among the ones found in literature gives better performance?
Elisa: if you try to use a C18 column with underivatized samples, you are going to see the same problem as the Reisdorph/Armstrong paper that Robin linked. The polar amino acids just will not bind effectively. I haven't read the whole paper carefully to see if they addressed the issue, but it's obvious from the column and flowrate they used (50x2.1 mm, 1.8 um, 0.2 mL/min) that the first batch of amino acids to elute (1-4 min) are basically in the void. I personally would not depend upon such chromatography to give reliable results: even with fully matched internal controls the matrix effects are going to be huge.
You can either derivatize or use a different column. I really like the Cogent Diamond Hydride: they've changed their website so it's harder to find the app notes, but they have one that shows all the amino acids. Here's a partial app: http://kb.mtc-usa.com/article/AA-00756/90/
To get back to your original question about extraction, there are about 10 recent papers regarding different metabolite extraction protocols from live cells (maybe start with Siuzdak). Coffee should be easier since you don't really have to worry about quenching. For this application I think you would be fine with any of the organic extract methods, although I'm partial to acetone (stepwise, followed by cooling).