Simulations explored the significance of formation water salinity and investigated approaches to simulating cement degradation in the presence of brines. Both the 'ion pairing and complexing' and 'specific ion interaction' ('Pitzer') approaches were used. Each one successfully reproduced the results of a cement carbonation experiment (covering 9 days) and field observations of cement degradation from the 'SACROC' site (covering 30 years), but gave increasingly differing results over progressively longer timescales. It is suggested that simulations of long-term seal performance in boreholes in contact with highly saline formation waters should adopt the Pitzer approach. The results illustrate that, on its own, matching simulation results and observed well seal behaviour over the short period for which data are available (a few tens of years) is an unreasonable basis on which to build confidence in long-term (>102 years) modelled seal performance.