The aim of this study was to develop and validate limited sampling strategies for accurately predicting 12-hour area under the concentration-time curve (AUC(0-12 h)) to provide a practical method for more precise therapeutic drug monitoring of cyclosporine A in stem cell transplant patients. Steady-state cyclosporine blood concentrations were measured within a dosing interval (12 hours post administration) in 35 allogeneic bone marrow transplant patients receiving 238 mg (±117 mg) twice-daily dose of cyclosporine. Limited sampling strategies were developed by multiple linear regression analysis of relationship between cyclosporine A full AUC(0-12 h) values and different combinations of preselected blood concentrations. Validation of the estimating equations was done by a bootstrap-like cross-validation method. Cross-validation results showed that cyclosporine AUC(0-12 h) could be estimated using either 2 or 3 samples within the first 4 hours after drug administration with good accuracy and precision (absolute prediction error of less than 6.2%). The number of estimated area under the drug concentration-time curves within 15% of observed values was greater than 26 (74%) for models used predose concentration with either c(2h) and c(4h) or both. Most of the previously reported single-sample models showed a systematic error in predicting AUC(0-12 h). Although a statistically significant difference in precision of prediction was seen between 3-sample model using c(0), c(2h), and c(4h) and 2-sample models (c(0), c(2h) or c(0), and c(4h)), such a difference (2%) could not be of clinical importance. Other 2-sample estimating equations (models using c(2h) with either c(6h) or c(10h)) with the same degree of precision appear to be less feasible clinically. Cyclosporine AUC(0-12 h) in bone marrow transplant patients could be estimated using 2 or 3 samples within the first 4 hours after drug administration with good accuracy and precision.