Background: Despite good adherence to supervised endurance training, a significant percentage of individuals (up to ~30%) shows no change in peripheral insulin sensitivity (SI) and some even demonstrate an adverse response [1]. The molecular mechanisms underlying this heterogeneous ability to improve SI through regular exercise are currently not well understood, but are likely to include a substantial genetic component [2]. Objective: to produce a molecular classifier that predicts SI training response guided by a multi-omics analysis framework. Methods: Peripheral SI was measured [intravenous glucose tolerance test] before and after a standardized 20-week endurance training programme [3 times/wk] in 478 healthy Caucasians from the HERITAGE Family Study [3]. Illumina HumanCNV370-Quad v3.0 BeadChips were genotyped. Affymetrix U133+2 arrays were used to quantitate gene expression levels from baseline limb muscle biopsies of a subset of participants (N=52). Results: The functional GWAS analysis identified several calcium signalling-related pathways associated with SI improvement (‘Cardiac muscle contraction’ being the most enriched; FDR<0.001). Noteworthy, multiple SNPs in close proximity to genes in the calcium signalling pathway were nominally associated with basal mRNA abundance (p<0.01). Furthermore, the mRNA expression levels of the calcium signalling pathway as a whole were differentially expressed at baseline between individuals with a high and low potential for improving their SI. We next reasoned that calcium signalling might regulate candidate transcription factors (TFs) of this pathway as a result of SNP variants affecting gene expression. Interestingly, the gene targets of the MEF2 TF family was positively associated with dSI (FDR<0.001), implying that individuals exhibiting high responsiveness of SI to training have an overall higher basal expression of genes co-regulated by MEF2. siRNA has previously been used on differentiated C2C12s to define the global transcriptional signature associated with MEF2 knockdown [4]. Genes down-regulated by knockdown of the MEF2A isoform (n=828 genes), which overall relate to ‘muscle function’, was highly enriched amongst the most positively associated genes to dSI in HERITAGE (FDR<0.001). Such in vitro validation prompted us to ask whether a robust multivariate regression model could be developed linking the basal mRNA abundance of MEF2A interacting gene targets to dSI on a continuous scale. The most predictive model (HDAC4, CAMK2D, CAMK2G) was able to explain nearly half (48%) of the variance of dSI in the HERITAGE sample. Importantly, the response predictor was successfully validated in an independent training cohort [5]. Conclusion: Combining data from genomics and transcriptomics analyses helped identify an RNA expression signature in resting muscle that is predictive of dSI.