Yeast gene expression profiles are analyzed with self-organizing map (SOM)-based exploratory clustering and information visualization methods. Each profile describes the expression of the other yeast genes after one gene has been removed or its function has been blocked. Information about the relationships of the function of the genes is hidden in the relationships between the profiles. We demonstrate that the clusters found by the SOM are closely related to the clusters found earlier by hierarchical clustering. The advantage of the SOM is the intuitive visualization of the similarity relationships and cluster structures within the data set. We additionally compare different metrics for the functional classification of the genes.