Microarrays have attracted tremendous interest among biologists. However, questions have been raised regarding the reproducibility between experiments performed in different laboratories using different platforms. Here, we discuss these problems and reach the following conclusions. First, the reproducibility between different platforms of arrays is low, but bioinformatics may permit compensation at least among oligonucleotide microarrays. Second, it is hard to interpret microarray results generated using mixed cell populations. Hierarchical clustering may be applied to identify whether upregulated transcripts in an inflammatory tissue were caused simply by an increased number of inflammatory cells.