Microarray technology provides a quantitative assessment of the number of gene transcripts gene using a high-throughput hybridization assay. Reliable detection of gene expression therefore requires reliable design of probes used in the hybridization assay. It is noted that microarray gene expression measurements are often characterized by variability, even among a series of technical replicate arrays. Therefore, sequence verification, used as a low-level filter to exclude probes exhibiting sequence inaccuracies, has previously been shown to reduce gene expression variability.Building on this work, the effects of sequence- and annotation-based filtering methods were quantified, and shown to be effective in reducing microarray variability among a set of technical replicates. Further, appropriate thresholds for filtering are recommended. A significant interaction in an analysis of variation model was found when a combination of sequence- and annotation-based filtering methods were explanatory factors, suggesting the use of the combination of filtering methods might be most beneficial.