The variation in human tumour pH values is large. The aim of this study was to analyse the reasons for these large variations and to determine whether tumour pH can be predicted on the basis of any easily measured parameter. One hundred and five determinations of tumour pH were performed in various human tumours, using the Philips C 902S tissue pH electrode. No correlations were found between the tumour pH and the tumour histology, degree of differentiation, tumour size, patient age or treatment history, and whether or not the tumour was ulcerated. However, tumour pH was significantly lower in primary tumours than in lymph node metastases. Tumours at their primary site (primary, recurrent or residual) were also more acid than distant metastases. The vascular disruption caused by the measuring technique was found to be acceptable.