The paper describes a system of cluster analysis for genome-wide expression data from DNA microarray hybridization, using standard statistical algorithms to arrange genes based on similarity in gene expression patterns. The output is displayed graphically, providing an intuitive representation of the clustering and underlying expression data for biologists. The authors found that clustering gene expression data in *Saccharomyces cerevisiae* groups genes with known similar functions together, and a similar trend was observed in human data. This approach allows for the interpretation of patterns in genome-wide expression experiments as indicators of cellular processes and can provide leads to the functions of poorly characterized or novel genes. The method combines hierarchical clustering with a graphical representation of the primary data, using color to reflect the original experimental observations. The results demonstrate that the clustered images contain large contiguous patches of color representing groups of genes with similar expression patterns over multiple conditions, which are not artifacts of the clustering procedure. The authors also discuss the biological significance of these patterns and the potential for applying similar methods to other large data sets.The paper describes a system of cluster analysis for genome-wide expression data from DNA microarray hybridization, using standard statistical algorithms to arrange genes based on similarity in gene expression patterns. The output is displayed graphically, providing an intuitive representation of the clustering and underlying expression data for biologists. The authors found that clustering gene expression data in *Saccharomyces cerevisiae* groups genes with known similar functions together, and a similar trend was observed in human data. This approach allows for the interpretation of patterns in genome-wide expression experiments as indicators of cellular processes and can provide leads to the functions of poorly characterized or novel genes. The method combines hierarchical clustering with a graphical representation of the primary data, using color to reflect the original experimental observations. The results demonstrate that the clustered images contain large contiguous patches of color representing groups of genes with similar expression patterns over multiple conditions, which are not artifacts of the clustering procedure. The authors also discuss the biological significance of these patterns and the potential for applying similar methods to other large data sets.