Presented at the 2016 Ocean Sciences MeetingOcean salinity is an integrator of changes in the water cycle, reflecting the exchange of freshwater between the ocean and various components of the climate system. In this study we will use the water mass transformation framework to infer water cycle changes from salinity changes in observations and CMIP5 models. The distribution of water in salinity coordinates is set by the stretching effect of the water cycle and the collapsing effect of mixing. A simple model is used here to describe the relationship between the width of the distribution, the water cycle and mixing, the latter being characterized by an e-folding timescale. In both observations and CMIP5 models we find that the water cycle maintains a salinity distribution in steady state with a mixing timescale of order 50 years. We show that observed changes in the width of the salinity distribution reveal an amplification of the water cycle of ~2.5% from 1950 to 2010. Both observations and models agree on a water cycle amplification of 5-6% per degree Celsius of surface temperature rise. CMIP5 models tend to be more diffusive than observations, potentially explaining why they display weaker surface salinity pattern amplification in response to warming. Our results show moderately weaker amplification than reported previously likely due to our inclusion of sub-surface salinity observations and careful treatment of net global fresh water input. The distribution of water in temperature-salinity coordinates is also considered here to investigate regional patterns of water cycle change.