The paper compares the results of two studies of salt damage at the Wakefield Tower, Tower of London. The first, in 1993, was based on semi-quantitative salt analysis and hygroscopicity measurements. The second was based on quantitative analysis and a thermodynamic model, ECOS. Both studies aimed to predict environmental conditions that would minimise salt damage to the interior stonework. Their predictions are markedly different, and this paper examines the reasons for the differences. It is shown that hygroscopicity alone is not a reliable indicator of crystallisation from mixed salt solutions.