The interstitial fluid pressure (IFP) and the proton spin-lattice and spin-spin relaxation times (T1 and T2) of some experimental tumours have been shown to be related to tumour water content. These observations have led to the hypothesis that magnetic resonance imaging (MRI) might be a clinically useful non-invasive method for assessment of tumour IFP. The purpose of the work reported here was to examine the general validity of this hypothesis. R-18 human melanoma xenografts grown intradermally in Balb/c nu/nu mice were used as the tumour model system. Median T1 and T2 were determined by spin-echo MRI using a 1.5-T clinical whole-body tomograph. IFP was measured using the wick-in-needle technique. No correlation was found between tumour IFP and fractional tumour water content. Moreover, there was no correlation between median T1 or T2 and IFP, suggesting that proton T1 and T2 values determined by MRI cannot be used clinically to assess tumour IFP and thereby to predict the uptake of macromolecular therapeutic agents.