What would be the advantage(if any) of a system suitability consisting of 5 injections from ONE standard? I have noticed that newer methods coming into the lab I work in have moved away from system suit's consisting of 3 injections of each of the two standards, which I believe to be a more rigorous test of a systems performance. Although the %RSD has been lowered from 2% to 1.5%, I do not understand the reason why, especially since standards must be made in duplicate and BOTH standards must be used in bracketing the samples in the run.

If both standards are used in the actual analysis, then methinks that both standards should be used in system suit (after all, preparing the standard is part of the method; by only using one standard, you are avoiding measuring one source of variability; no wonder the precision *looks* better).

On the other hand, this is partly a scientific issue and partly a regulatory issue. If you have an SOP in place that says using a single standard for system suit is OK, then you have your tail covered. Likewise, if the method was validated with a system suit based on one standard, then by definition, there is no problem running it this way.

In the absence of an SOP or an explicit direction in the manual, I'd use the two standards.

If you estimate a pooled s.d. from both standard samples, you will have 3+3-2=4 degree of freedom for the confidence interval (for the true s.d.) Those are the same d.f. for the estimate of five replicate injections. It's quite clear that preparing two standard solutions (by weighting the reference compound) is better than one standard, even performing five injection, because you can calculate the ratio of response factors [(mean_area1.W2)/(mean_area2.W1)] which should be between 0.99 and 1.01 if the standard solution were correctly prepared.
That's why I think it would be better 3 inj x 2 std. than 5 inj. (1st std.) + 2 (2nd std.). The same inf. with one less injection.