Most accurate method? None as all are very good. *The operator's experience and method selected will really be the limiting factor in how accurate any of these methods are.

GC/FID, GC/MSD, LC/DAD and LC/MS are all very accurate techniques to use for the determination and quantification of caffeine in many sample types. However, for a clinical sample of this type, I would recommend a liquid based method such as HPLC/DAD or HPLC/DAD/MS.

BTW: ALL of these methods are really easy to setup and run.

Never consider a single or multi-wavelength detector for any HPLC method development work. Use only a full scanning capable diode array detector (AKA, DAD or PDA) for HPLC method development work. Diode array detectors provide you with the ability to 'see' many things in the 3rd dimension (1st dim retention, 2nd dim Lamda Max, 3rd dim full spectra). This allows you to qualitatively discern other compounds from your primary and detect impurities and co-eluting peaks. A Mass detector coupled to the DAD adds the ability to 'see' any ionized forms (and possible fragments too) which can aid in identification.

If money is a critical issue, a GC/FID/MSD system can also be used and should cost far less than any HPLC/DAD/MS system, plus they are much easier to use (far less training required). Easy to see peak at 194 with nice fragmentation pattern by EI.

Thank you Bill Letter for spending some time answering my question. Your answer was very enlightening! HPLC does look like the most used method. Also I really appreciate your suggestions regarding the "detector" method

Hi Ricardo,

In addition to the remarks made by Bill, I'd like to add that the accuracy of the final determination also greatly depends on the procedure that you're planning to use to extract caffeine from blood plasma. Do you have any idea what the characteristics (such as recovery and reproducibility) of that method are? If these are poor, no LC or GC will help you get accurate results, regardless of the detector used.

In that sense, it might be worthwhile to consider using an internal standard, such as stable isotope-labeled caffeine, when you're thinking of using MS as a detection method.

Hope this helps.