Abstract Background: Although TRIzol is widely used for preservation and isolation of RNA, there is suspicion that prolonged sample storage in TRIzol may affect array-based gene expression profiling (GEP) through premature termination during reverse transcription. Methods: GEP on Illumina arrays compared paired aliquots (cryopreserved or stored in TRIzol) of primary samples of multiple myeloma (MM) and acute myeloid leukemia (AML). Data were analyzed at the “probe level” (a single consensus value) or “bead level” (multiple measurements provided by individual beads). Results: TRIzol storage does not affect standard probe-level comparisons between sample groups: different preservation methods did not generate differentially expressed probes (DEP) within MM or AML sample groups, or substantially affect the many DEPs distinguishing between these groups. Differences were found by gene set enrichment analysis, but these were dismissible because of instability with permutation of sample labels, unbalanced restriction to TRIzol aliquots, inconsistency between MM and AML groups, and lack of biological plausibility. Bead-level comparisons found many DEPs within sample pairs, but most (73%) were <2-fold changed. There was no consistent evidence that TRIzol causes premature reverse transcription termination. Instead, a subset of DEPs were systematically due to increased signals in TRIzol-preserved samples from probes near the 5′ end of transcripts, suggesting better mRNA preservation with TRIzol. Conclusions: TRIzol preserves RNA quality well, without a deleterious effect on GEP. Samples stored frozen with and without TRIzol may be compared by GEP with only minor concern for systematic artifacts. Impact: The standard practice of prolonged sample storage in TRIzol is suitable for GEP. Cancer Epidemiol Biomarkers Prev; 19(10); 2445–52. ©2010 AACR.