Microarray-based gene profiling of laser-assisted microdissected tissues or clinical biopsies is still a challenge since the amount of total RNA in such samples is limited and amplification of RNA is mandatory. Representative amplification of mRNA is highly dependent on the reverse transcription reaction, which is error prone, and on the number of amplification cycles. To improve the accuracy of RNA amplification, we optimized, combined, and tested different amplification strategies for Affymetrix oligonucleotide array hybridization. We demonstrate that different protocols differ significantly in quality of mRNA amplification. To demonstrate the accuracy and reproducibility of our optimized protocol in a clinical setting, we analyzed total RNAs from laser-assisted, microdissected cells of human prostate tissues. On the basis of these results, we recommend a standard reverse transcription reaction for small-sample-transcriptome profiling experiments as part of the Minimal Information about a Microarray Experiment (MIAME) set of standards.