Large amounts of hydrocarbon reserves are trapped in naturally fractured reservoirs which arechallenging in terms of accurate recovery prediction because of their joint fabric complexity andlithological heterogeneity. Canada, for example, has over 400 billion barrels of crude oil in fracturedcarbonates in Alberta, most of this being bitumen of viscosity greater than 106 cP in the GrosmontFormation, which has an average porosity of about 13-15%. Thermal methods are the most commonexploitation approaches in such viscous oil reservoirs which, in the case of steam injection, are associatedwith up to 275-300°C temperature changes, leading to considerable thermoelastic expansion. Thistemperature change, combined with pore pressure changes from injection and production processes, leadsto massive effective stress variations in the reservoir and surrounding rocks. The thermally-induced(thermoelastic) stress changes can easily be an order of magnitude greater than the pore pressure effectsbecause of the high intrinsic stiffness of the low porosity limestone and bounding strata. Study of thesestress-pressure-temperature effects requires a thermo-hydro-mechanical (THM) coupling approach whichconsiders the simultaneous variation of effective stress, pore pressure, and temperature and theirinteractions. For example, thermal expansion can lead to significant joint dilation, increasing themacroscopic, joint-dominated transmissivity by an order of magnitude in front of and normal to thethermal front, while reducing it in the direction tangential to the heating front. This leads to stronginduced anisotropy of transport processes, which in turn affects the spatial distribution of the heatingarising from advective heat transfer.