Background The need to perform microarray experiments with small amounts of tissue has led to the development of several protocols for amplifying the target transcripts. The use of different amplification protocols could affect the comparability of microarray experiments. Results Here we compare expression data from Pinus taeda cDNA microarrays using transcripts amplified either exponentially by PCR or linearly by T7 transcription. The amplified transcripts vary significantly in estimated length, GC content and expression depending on amplification technique. Amplification by T7 RNA polymerase gives transcripts with a greater range of lengths, greater estimated mean length, and greater variation of expression levels, but lower average GC content, than those from PCR amplification. For genes with significantly higher expression after T7 transcription than after PCR, the transcripts were 27% longer and had about 2 percentage units lower GC content. The correlation of expression intensities between technical repeats was high for both methods (R2 = 0.98) whereas the correlation of expression intensities using the different methods was considerably lower (R2 = 0.52). Correlation of expression intensities between amplified and unamplified transcripts were intermediate (R2 = 0.68-0.77). Conclusion Amplification with T7 transcription better reflects the variation of the unamplified transcriptome than PCR based methods owing to the better representation of long transcripts. If transcripts of particular interest are known to have high GC content and are of limited length, however, PCR-based methods may be preferable.