Application of metabolomics techniques to plant physiology is now considerable, and LC-MS is often being used for non-targeted, semi-quantitative analysis of effects caused by mutations or environmental conditions. However, examination of signalling metabolites like hormones require absolute rather than semi-quantitative quantitation, since their effect in planta is strongly dependent upon concentration. Further, plant hormones belong to different chemical classes and thus simultaneous quantitation remains highly challenging. Here we present an LC-MS method that allows the simultaneous absolute quantitation of six hormone families as well as selected phenolics. The technique requires solid phase extraction with a sulfonated cation exchange phase before analysis, and use calibration curves instead of isotopically labelled standards, which are indeed not commercially available for many hormonal molecules. The use of the total signal (including adducts) rather than a single quantifying mass appears to be crucial to avoid quantification errors because the ion distribution between adducts is found to be concentration-dependent. The different hormones considered appear to have contrasted ionisation efficiency due to their physical properties. However, the relatively low variability and the satisfactory response to standard additions show that the technique is accurate and reproducible. It is applied to Arabidopsis plants subjected to water stress, using either the wild-type or lines with altered NAD biosynthesis causing changes in salicylate signalling and phenylpropanoid levels. As expected, analyses show an increase in abscisic acid upon water stress and a consistent modification of phenolic compounds (including salicylate) in mutants.