Dear Bin,

all molecules that undergo ionisation process can be measured by MS-based strategies. The term "small molecules" is a bit vague, do you mean with Mw

 Thanks. But in practice, it is difficult to obtain the isotopically labelled derivatives and to judge whether the structural analogues show the similar ionisation efficiency or not. 

If you don't have reference standard, how do you know you got the right compond. For MS/MS based instrument, you will need retention time and MS spectra or precursor/product ion ratio to confirm the identity of the unknown. However, if you don't have ref standard and you want to know if you have the specific analyte, e.g. you inject serum and get the chromatogram by ESI, you may use the accurate mass (ToF or Orbitrap) and determine the accurate mass. Say cocain has mass 200.1234, you may ask the instrument to plot a chromatogram of mas 200.1234. It will screen out lots of compounds and give you a limited number of peak that has mass 200.1234. Once you see the peak, then you can run MS/MS or full scan MS/MS to see the fragmentation and check if it might match with the literature. Again, at the end of the day, you still need ref to confirm the presence of the analytes.

Check this article for the application of accurate mass non-target ID

http://www.absciex.com/Documents/Downloads/Literature/tripletof_non_target_screening_identification_AB%20SCIEX.pdf

Can we say that reliable quantification of unknown anlytes is impossible? But for some structually similar coumpounds such as drug metabolites, their concentrations may be measured by quantifying their parent drugs.

If you don't have standard, how would you measure the amount accuratly? Yes, you can guess by measuring the response against something with similar structure and you know the quantitity, again it is the estimate. You can say the unknown is approximately 10 times the amount of the compound you have in the same chromatogram. You don't know if the response is 1:1 ratio. You just do the best as you can for what you have.

It may be used for relative but not for absolute quantitation.

Is it better than LC-MS to use UV absorption based on their similar extinction coefficients? 

You have no idea about your unknow, how you can assume that? Estimate is the estimate. Your guess is as good as mine. You are trying to use a bicycle to go everywhere at 20 mile/hour, something has to give.  At least you can use all to tools in your lab to do the job, at one point you will need the right tool to do the right job.