Flow back or produced water from shale gas hydraulic stimulation has been reported, most notably in the USA, to contain high concentrations of dissolved metals such as Na, As, Cd, or U which may originate from formation water or fluid-mineral reactions. An understanding of the fluid-mineral geochemical reactions occurring can be gained from constrained experimental studies combined with validated predictive kinetic geochemical modelling. Shales of the Cooper Basin Roseneath-Epsilon-Murteree (REM) sequence are unconventional gas targets in Australia. Two core samples from the Roseneath shale and Murteree shale were geochemically characterised in detail. Both shales contained 55-57% quartz, 14-18% illite/muscovite, 3-5% kaolinite and 15-17% (Mg)-siderite. The Roseneath shale sample from 3266 m contained relatively more sphalerite and pyrite, with a higher V, Cr, Cu, Zn and Pb content. The Murterec shale from 3497 m, contained relatively more siderite and also ankerite, and had a higher Fe, Mn, and Mg content. On experimental reaction with water and traces of air at 75 degrees C and 200 bar of N-2, siderite was dissolved in both cases, and ankerite was dissolved from Murteree shale. Sphalerite, pyrite, and siderite oxidative dissolution reduced the solution pH to similar to 3 during reaction of the Roseneath shale, with open pores and cubic Fe-rich precipitates formed. Higher concentrations of dissolved Cr, Co, Ni, Cu, Zn, Cd, and U were released from the Roseneath shale than the Murteree, though most released concentrations were < 10% of the total amount measured in the core. Ankerite in the Murteree shale initially buffered solution pH to 5.5, but subsequently pH decreased to 4. Spherical Fe-oxides precipitated during reaction of the Murteree Shale Cr signatures sequestering metals. Higher concentrations of dissolved Mn, Mg, Ca, Na, Sr, Mo and Hg were initially mobilised from the Murteree shale, with Na, Mo and Hg concentrations subsequently decreasing. Geochemical kinetic modelling of the experiments was performed, using experimental data to estimate mineral reactive surface areas. Mineral reactive surface areas in geochemical models were increased to 1000 cm(2)/g for clays, and 10-30 cm(2)/g for carbonates to approximate the experimental water chemistry. Models confirmed the dissolution of siderite, pyrite, sphalerite, and for Murteree shale also ankerite. Fe-oxide, siderite, or sulphide precipitation was predicted. The amount and reactivity of carbonate minerals and the presence of sulphides affect the fluid acidity and dissolved metal content. Metal concentrations in produced water will be site specific partly depending on the mineralogy of the formation and chemistry of injected fluids. Understanding the geochemical evolution of fluids and accurately predicting them via combined experimental and validated model studies could lead to improved upscaled predictions of fluid chemistry or strategies for treatment. The precipitation of Fe-oxides could clog pore throats or fractures, and has the potential to decrease permeability and long-term gas production.