We show here that the amount of labeled cDNA and its specific activity can play a significant role in the quantitation of microarray experiments. Standard reverse transcription of 2 microg total bacterial RNA with concomitant incorporation of cyanine dye-conjugated nucleotides did not produce enough label for optimal hybridization results in our Mycobacterium tuberculosis DNA microarray. Therefore we turned to an alternative labeling method using the incorporation of aminoallyl nucleotides followed by conjugation to Cy-dye. The method allows up to 10 fold more label to be produced, and at higher specific activity. In particular, more transcripts can be detected and variability between replicate features can be reduced by using more labeled cDNA. We show that optimizing the labeling protocol is a critical element in conducting microarray experiments and obtaining reproducible and interpretable data.