Creatinine-based methods for measuring glomerular filtration rate (GFR) were compared with the plasma clearance of 99mTc DTPA as an accurate index of GFR in patients prior to and during cisplatin therapy. 2. Measured creatinine clearances using either unsupervised 24 h or supervised 4 h collections correlated less well with the reference GFR than did predicted creatinine clearances derived from serum creatinine using formulae or a nomogram. 3. Predicted creatinine clearance appears to be at least as sensitive as measured creatinine clearance for screening GFR in patients receiving cisplatin. However, owing to the inherent inaccuracy in any creatinine-based method, if the prediction is within 20% of the value at which dose reduction of cisplatin is considered, an isotopically determined GFR should be carried out to optimise dosage adjustments.