Due to the high cost of microarray experiments, investigators typically select designs with biological rather than technical replicates. Therefore, it is essential that the quality of RNA hybridized to the microarray meets certain standards. The process of transcription begins with reverse transcriptase binding at the 3' end of a gene and continuing toward the 5' end. However, transcription generally does not continue to completion. That is, reverse transcription typically drops off before reaching the 5' end. Affymetrix GeneChips includes probe sets which interrogate both the 3' end and the 5' end for selected control genes to assess quality of transcription. The MAS 5.0 software estimates the 3'/5' ratio after the PM and MM probes have been summarized into a probe set expression measure. Unfortunately, inherent to all probe set expression summary methods is that the 3' and 5' probe sets of interest are only represented on the GeneChip once. This leads to the unfortunate consequence of inadequate replications for variance calculation. The methodology proposed uses the pixel level intensities to increase the number of observations per probe set in order to obtain a better estimate of the 3' to 5' ratio. Since there is an inherent hierarchical structure to GeneChip data, where pixels are nested within probes and probes are nested within probe sets, RNA degradation will be assessed by fitting mixed effects ANOVA models to estimate the 3' to 5' ratio treating probe set as a fixed effect, while treating pixel level and PM level data as random effects. This enables the construction of confidence intervals about the estimated ratio. The estimated confidence interval will more appropriately indicate whether the RNA was of sufficient quality rather than judging quality based on the ratio being below an arbitrarily selected threshold. Results from HG-U133A GeneChips will be presented.