As well as Duncan’s point (see post below), Sherman, Molloy and co-workers, also in Proteomics (vol 9), believe that monitoring the transition of a single precursor to product ion is not specific enough to define a unique peptide anyway. This is because several peptide precursors may have equivalent mass, or may appear to have, due to inadequate mass resolution of the MS/MS instrument.
Moreover, particularly in complex samples, MS/MS level product ions derived from different precursors may fall within the same m/z window producing overlapping signals. As a result, determining the quantity of the original peptide, and hence protein, is fundementally flawed.
In essence, Molloy and co are suggesting that targeting unique PEPTIDES is not enough for SRM, you need unique TRANSITIONS. And not only that, you also need to have MULTIPLE TRANSITIONS for each protein in order to verify the accuracy of the quantification across more than one target.
The situation is not so bleak, however, because Duncan et al suggest that, in reality, the chances of two peptides having the same mass in a specific experiment is actually quite low, and that HPLC separation is pretty good at separating even the most closely related peptides, so there is little chance of near-isobaric peptides entering the instrument simultaneously in the first place. However, as ever, if you want to be really sure you are measuring what you think you are – take a consensus approach.