RNA amplification methods have been used to facilitate making probes from small tissue samples for microarray studies. Our original amplification technique relied on driving the first reverse transcription with oligo(dT) with a T7 RNA polymerase promoter (T7dT) on the 5' end, and subsequent transcriptions with random 9mers with a T3 RNA polymerase promoter (T3N9). Thus, initially, poly(A)(+) RNA is amplified. This creates a potential problem: amplifications based on oligo(dT) priming could be sensitive to RNA degradation; broken mRNA strands should give rise to shorter cDNAs than those seen when intact templates are used. This would be especially troublesome when targets other than those corresponding to the 3' ends of transcripts are printed on an array. To solve this problem, we elected to prime cDNA synthesis with T3N9 at the beginning of each amplification cycle. Following two rounds of amplification, the resulting probes were comparable to those obtained with our original protocol or the Arcturus RiboAmp kit. We show below that as many as four rounds of amplification can be performed reliably. In addition, as predicted, the method works well with degraded templates.