This dissertation reports on studies on the role of the mean state in modulating climate variability at interannual and decadal time scales. In the atmosphere, the nonlinear superposition of mean flow and anomalous flow has important implications for many phenomena associated with variables that are nonlinear by definition, such as the vertical wind shear and surface wind speed. In the first part of this dissertation, the influence of mean flow and anomalous flow on vertical wind shear variability is studied in observations and numerical model simulations. At interannual timescales, the ENSO-shear relationship is compared between observations and numerical model simulations. It is shown that there is strong influence of mean flow on the ENSO-shear relationship. For same anomalous flow, different mean flows could give rise to a different ENSO-shear relationship. The nonlinear superposition of mean flow and anomalous flow also helps explains the dipole mode of tropical Atlantic vertical wind shear variability seen in observations and models, which implies opposite variation of vertical wind shear over the two sides of the tropical Atlantic. This has important implications for predicting phenomena such as Atlantic hurricanes, whose variations are modulated by vertical shear variability. The dissertation also addresses the role of the mean surface wind in decadal variability and predictability, as manifested through the Wind-Evaporation-SST (WES) feedback. The nonlinear superposition of anomalous surface wind on the mean trade wind can give rise to a positive WES feedback, which can amplify tropical climate variability. To study this feedback, we carried out ensembles of decadal climate predictions using the CAM3 atmospheric model coupled to a slab ocean model (CAM3- SOM) with prescribed ocean transport and simple extrapolative prescriptions of future external forcings. Mechansitic sensitivity runs using the CAM3-SOM were also carried out, where the WES feedback was switched off by prescribing climatological surface wind. Results suggest that switching off the WES feedback enhances the prediction skill over some regions, especially over the eastern tropical Pacific, by increasing the signal- to-noise ratio. To address the issue of cold bias noted in the decadal prediction experiments, we carried out additional sensitivity experiments where we used an adaptive formulation for the prescribed oceanic heat transport (Q-flux) in the slab ocean. The results from these experiments demonstrate that the mean oceanic heat transport plays a crucial role in influencing decadal predictability, by helping improve predictions of the trend component of decadal variations.