This study was undertaken to analyse the relationship between total calcium (TCa) and ionized calcium (ICa) in patients with cancer, and to assess the clinical value of routine measurements of ICa in these patients. Serum TCa, ICa, albumin, proteins and creatinine were measured in 188 adult patients with solid malignant tumours. Most of them were out-patients, the Karnofsky score being 80 or above in 67%. The correlation coefficient between ICa and TCa was 0.85 (P less than 0.001) and did not improve after correcting TCa for protein concentration with several published formulae. Although TCa measurements had a global diagnostic accuracy (percent of patients correctly classified) of 90%, they failed to identify a substantial proportion of patients with increased levels of ICa (57% for uncorrected TCa, and 27-57% for protein-corrected TCa). However, the finding of slightly increased ICa levels did not seem to predict the development of frank hypercalcaemia and did not impair the prognosis. According to these results, the routine measurement of ICa in unselected patients with cancer has no clinical usefulness.