Gene expression analyses using spotted cDNA microarrays typically require relatively large quantities of total RNA (up to 100 microg) or polyA+RNA (1-5 microg). However, samples obtained by microdissection, patient biopsies, or embryonic samples often are small and yield an insufficient amount of RNA. Methods such as linear RNA amplification by in vitro transcription (IVT) or cDNA amplification by PCR are currently being used to circumvent these limitations. In the present study, labeled probes from mouse liver and kidney were generated with two amplification methods and were analyzed in terms of reproducibility of intensity values from repeated experiments. In addition, the reliability of differential gene expression detection among the different types of amplified and non-amplified probes was assessed. Data derived from IVT-amplified RNA, as well as from PCR-amplified cDNA probes were reproducible with correlation coefficients of 0.89 and 0.91, respectively. 88-92% of the strongly differentially expressed genes detected with non-amplified probes were also detected as being at least two-folds differentially expressed with the amplified probes. Both the PCR-amplified probe and the IVT-amplified probe were comparable in reproducibility and reliability.