Analytical performance of stable isotope-labeled internal standardization (SIL-IS) and threshold accurate calibration (TAC) methods of matrix normalization are compared for quantitation of 51 drugs and metabolites (analytes) in urine with analysis by ultra performance liquid chromatography with tandem mass spectrometry (UPLC-MS-MS). Two SIL-IS methods of analysis were performed, one method using analyte-specific internal standardization (ASIL-IS) and another method using a shared stable isotope from another analyte for internal standardization (SSIL-IS). Variance in inter-specimen matrix effect, without the use of a matrix normalization method, was studied by UPLC-MS-MS analysis of 338 urine donor samples and showed >200% variation in ion response for some analytes. Matrix normalization methods were evaluated for precision, accuracy, calibration, multi-matrix recovery and positive casework quantitation. Acceptable calibration and quality control criteria were achieved for all methods when calibrators and controls were prepared from the same urine matrix pool. Quantitative accuracy, determined by the addition of analytes to multi-donor urine pools at two concentration levels, resulted in acceptable percent relative standard deviation (%RSD) and bias for TAC and ASIL-IS methods. SSIL-IS method quantitations in analyte-supplemented donor pools revealed a %RSD ranging from 20% to 60% for >30% of the analytes and a method bias that ranged up to 87%, with a differential matrix effect on analyte and shared internal standard accounting for the imprecision and bias. Analyte quantitation in 162 authentic case samples showed close agreement for TAC and ASIL-IS methods, with greater variance in the SSIL-IS method. The study demonstrates effective matrix normalization by ASIL-IS and TAC methods and a matrix-caused bias in the SSIL-IS method.