The development and validation of a LC-MS/MS method is often performed using pooled human plasma, which may fail to account for variations in interindividual matrices. Since calibrator standards and quality control samples are routinely prepared in pooled human plasma, variations in the extraction recovery and/or matrix effect between pooled plasma and individual patient plasma can cause erroneous measurements. Using both pooled human plasma as well as individual healthy donor and cancer patient plasma samples, we evaluated the analytical performance of two classes of internal standards (i.e., non-isotope-labeled and isotope-labeled) in the quantitative LC-MS/MS analysis of lapatinib. After exhaustive extraction with organic solvent, the recovery of lapatinib, a highly plasma protein-bound drug, varied up to 2.4-fold (range, 29 – 70%) in 6 different donors of plasma and varied up to 3.5-fold (range, 16 – 56%) in the pretreatment plasma samples from 6 cancer patients. No apparent matrix effects were observed for lapatinib in both pooled and individual donor or patient plasma samples. The calibration curve range was 5 – 5000 ng/ml of lapatinib in plasma. Both the non-isotope-labeled (zileuton) and isotope-labeled (lapatinib-d3) internal standard methods showed acceptable specificity, accuracy (within 100 ± 10%), and precision (< 11%) in the determination of lapatinib in pooled human plasma. Nevertheless, only the isotope-labeled internal standard could correct for the interindividual variability in the recovery of lapatinib from patient plasma samples. As inter- and intra-patient matrix variability is commonly presented in the clinical setting, this study provides an example underscoring the importance of using a stable isotope-labeled internal standard in quantitative LC-MS/MS analysis for therapeutic drug monitoring or pharmacokinetic evaluation.