The impact of lower serum albumin levels on teicoplanin pharmacokinetics has not been previously determined. The authors assessed the relationship between total and free concentrations of teicoplanin in serum samples obtained from patients receiving teicoplanin therapy for Gram-positive bacterial infections. In addition, the authors determined the contribution of serum albumin concentrations to the unbound fraction of teicoplanin. One hundred ninety-eight serum samples were obtained from 65 patients undergoing routine therapeutic drug monitoring of teicoplanin. Free serum teicoplanin was separated by ultrafiltration, and total and free serum concentrations of teicoplanin were determined by a fluorescence polarization immunoassay. Regression analysis was then performed to build a prediction model for the free serum teicoplanin concentration from the total serum teicoplanin concentration and the serum albumin level using the first 132 samples. The predictive performance of this model was then tested using the next 66 samples. Free serum teicoplanin concentrations (Cf) (mug/mL) were predicted using a simple model constructed using total serum teicoplanin (Ct) (mug/mL) and albumin concentrations (ALB) (g/dL): Cf = Ct/(1 + 1.78 * ALB). This model could estimate free serum teicoplanin concentrations with a small bias and an acceptable error. The measured free level of teicoplanin will lie between 0.63 and 1.38 times the predicted concentration in 95% of cases. Serum albumin level plays a major role in the variability of the fraction unbound of teicoplanin. This model can reliably estimate free serum teicoplanin concentrations more easily than by using direct measurements.