The completion of the Human Genome Project 1 has made possible the comprehensive analysis of gene expression 2, 3, and cDNA microarrays are now being employed for expression analysis in cancer cell lines 4 or excised surgical specimens 5. However, broader application of cDNA microarrays is limited by the amount of RNA required: 50–200 μg of total RNA (T-RNA) and 2–5 μg poly (A) RNA 6. To broaden the use of cDNA microarrays, some methods aiming at intensifying fluorescence signal 7, 8, 9 have resulted in modest improvement. Methods devoted to amplifying starting poly (A) RNA 10, 11 or cDNA 12 show promise, in that detection can be increased by orders of magnitude. However, despite the common use of these amplification procedures 11, 13, 14, 15, 16, no systematic assessment of their limits and biases has been documented. We devised a procedure that optimizes amplification of low-abundance RNA samples by combining antisense RNA (aRNA) amplification 10 with a template-switching effect (Clonetech, Palo Alto, CA). The fidelity of aRNA amplified from 1: 10,000 to 1: 100,000 of commonly used input RNA was comparable to expression profiles observed with conventional poly (A) RNA-or T-RNA-based arrays.