Hi jcwagner,
Other than the autosampler (variation of injection volume, perhaps a need to perform "priming injections" of standard solutions that are more concentrated than the typical standard solutions you normally use, health of the needle and injection valve) and the pump (flow rate variability can cause peak area variability--think of the pump seals, check valves and in-line/valve filters), I'm not sure, other than the software and detector settings--these last two are thought to be okay as you've used them before with this method.
Section 13. of this reference may be helpful:
http://www.waters.com/webassets/cms/lib ... a20769.pdfPlease, these injections you're making, are they of standard and sample solutions both, or standard solutions only (n = 6) from the same vial (perhaps sorption to the vial?)? Sometimes also I have found that for trace analyses, the degree of imprecision with peak height is less than that of peak area--is it possible to alter the calibration of the method you're using if this can be demonstrated?