The number of RNA molecules per cell (transcriptome size) is highly variable, differing among and within cell types depending on cell size, stage of the cell cycle, ploidy level, age, disease state, and growth condition. Such variation has been observed at the level of total RNA, ribosomal RNA, messenger RNA (mRNA), and the polyadenylated fraction of mRNA, and these distinct RNA species can also vary in abundance with respect to each other. This variation in transcriptome size has been largely ignored or overlooked, and in fact, standard data normalization procedures for transcript profiling experiments implicitly assume that mRNA transcriptome size is constant. Consequently, variation in transcriptome size has important technical implications for such experiments, as well as profound biological implications for the affected cells and underlying genomes. Here, we review what is known about transcriptome size variation, explore how ignoring this variation introduces systematic bias into standard expression profiling experiments, and present examples of how such biases have led to erroneous conclusions in expression studies of sex chromosome dosage compensation, cancer, Rett syndrome, embryonic development, aging, and polyploidy. We also discuss how quantifying transcriptome size will help to elucidate the selective forces underlying patterns of gene and genome evolution and review the evidence that cells exert tight control over transcriptome size in order to maintain cell size homeostasis and to optimize chemical reactions within the cell, such that loss of control over transcriptome size is associated with cancer and aging. Thus, transcriptome size is an important phenotype in its own right. Finally, we discuss strategies for quantifying transcriptome size and individual gene dosage responses in order to account for and better understand this important biological phenomenon.