Large-scale gene expression profiling using microarray technology is often limited by the amount of tissue or cells available. A number of RNA amplification protocols have been published to overcome this problem. However, additional amplification steps can result in both a 3' bias and poor reproducibility for low abundance transcripts. We performed microarray experiments using HG-U133A GeneChip arrays to ascertain whether less than the recommended amount of RNA can be used, thus avoiding additional amplification steps. In a titration experiment, 2-10 mu g of total RNA from a single cryopreserved patient specimen was used to prepare biotinylated cRNA, and the recommended standard amount of 15 mu g of each preparation was used for hybridization. Statistical analysis using box plots, correlation coefficients, MvA plots, and concordance percentages revealed almost identical levels of gene expression, independent of the amount of RNA used for target preparation. Most importantly, there was no statistically significant difference when the concordance percentages for low abundance genes were compared, demonstrating that as little as 2 mu g of total RNA is sufficient to perform GeneChip analysis.