DNA microarrays represent an important new method for determining the compl
ete expression profile of a cell. In "spotted" microarrays, slides carrying
spots of target DNA are hybridized to fluorescently labeled cDNA from expe
rimental and control cells and the arrays are imaged at two or more wavelen
gths. In this paper, we perform statistical analysis on images of microarra
ys and show that quantitating the amount of fluorescent DNA bound to microa
rrays is subject to considerable uncertainty because of large and small-sca
le intensity fluctuations within spots, nonadditive background, and fabrica
tion artifacts. Pixel-by-pixel analysis of individual spots can be used to
estimate these sources of error and establish the precision and accuracy wi
th which gene expression ratios are determined. Simple weighting schemes ba
sed on these estimates are effective in improving significantly the quality
of microarray data as it accumulates in a multiexperiment database. We pro
pose that error estimates from image-based metrics should be one component
in an explicitly probabilistic scheme for the analysis of DNA microarray da
ta.