Carboplatin shares some of the therapeutic advantages of cisplatin, but without a significant incidence of the dose-limiting neurotoxicity and nephrotoxicity which is experienced with cisplatin. However, its use is associated with dose-limiting bone marrow suppression. Carboplatin is present in the blood as 3 distinct species. These are total platinum and 2 unbound species, carboplatin itself and a decarboxylated platinum-containing degradation product. The 2 main methods used to assay the unbound species are flameless atomic absorption spectrophotometry and high performance liquid chromatography. The first of these methods assays both unbound platinum species, the second is specific for carboplatin. Both unbound species have similar pharmacokinetic profiles for the first 12 hours post-dose. Carboplatin appears to have a linear pharmacokinetic profile over the doses used clinically and does not interact significantly with drugs that are used commonly in combination chemotherapy. The pharmacokinetics of carboplatin are adequately described by an open 2-compartment model with elimination from the central compartment. Its clearance is proportional to the glomerular filtration rate and the volume of distribution of the central compartment appears to correlate with extracellular fluid volume. The elimination half-life varies with renal function and is typically between 2 and 6 hours in patients with a normal glomerular filtration rate and may be as long as 18 hours in patients with impaired renal function. Relationships between systemic exposure to carboplatin, described as the area under the concentration-time curve (AUC), and both toxicity and response have been described. For toxicity the strongest evidence exists for a relationship between AUC and thrombocytopenia. To a lesser extent the relationship between AUC and neutropenia has also been described. Patients already treated with platinum analogues have been shown to develop a greater degree of myelosuppression from any given AUC. In addition, some evidence suggests a relationship between the shape of the concentration-time curve and myelotoxicity, where constant infusions appear less likely to cause myelosuppression on a mg/m2 dose administration basis. The relationship between AUC and response rate is not as clear, this may be related to the lack of studies describing both the dose and AUC of carboplatin. There appears to be a more clearly defined AUC-response relationship for ovarian cancer than for other malignancies, with an AUC of between 5 and 7 mg/ml.min being associated with the maximal response rate [located at the plateau on an AUC-response curve]. However, new data suggest that higher AUCs may lead to greater response rates. Data from testicular cancer also strongly supports an AUC-response relationship with an increased number of treatment failures with carboplatin AUCs < 5 to 6 mg/ml.min. Given the AUC-effect relationships described above a number of studies have been performed to develop models to describe the relationship between both dose and AUC and dose and platelet nadir. In adults, perhaps the most common method is that of Calvert which describes the relationship between dose and AUC. Paediatric formulas have also been described. More recently a number of limited sampling strategies have been proposed as well as Bayesian dose individualisation techniques.