The erythromycin breath test (EBT) is a putative probe of cytochrome P450 (CYP) 3A4 activity in vivo. Therefore, the EBT might prove useful for the individualisation of doses of drugs that have a low therapeutic window (for example the immunosuppressants or cytotoxics) and are metabolised by CYP3A4. However, there is a lack of consensus as to how the EBT should be used to predict total body clearance (CL), and the results so far have been largely disappointing. We argue that the required assumption that individuals produce 5 mmol of CO2/min per m2 at rest is one of the problems with the existing EBT, as the literature suggests significant variability and possible gender differences in this parameter. An examination of the EBT with a simple compartment model suggests that alternative parameters could be more useful in the prediction of CL. In particular, there is theoretical support for the use of the time-point at which breath radioactivity is maximal (tmax) as a correlate for CL. This is in agreement with our recent study of the pharmacokinetics of erythromycin in patients with cancer.