This paper proposes a method for improving the assessment of differential expression in microarray experiments by utilizing within-array replicate spots. The method estimates the correlation between replicate spots and uses this information to improve the precision of gene-wise variance estimates. This leads to more accurate inference methods for identifying differentially expressed genes. The method is validated using data from a microarray experiment involving calibration and ratio control spots in conjunction with spiked-in RNA. The results show that the common correlation method results in substantially better discrimination of differentially expressed genes from those which are not compared with simply averaging the replicate spots. The method may be combined with empirical Bayes methods for moderating the gene-wise variances between genes. The method is implemented in the limma software package for R, available from the CRAN repository. The methodology is applicable to a variety of microarray experiments and has been shown to be effective in experiments with small sample sizes. The method is particularly useful for experiments with poor quality data or with a limited number of arrays. The common correlation method is shown to have greater power in detecting differential expression while producing no more false positives on average. The method is also shown to be effective in experiments with very few arrays or with poor quality data. The method is implemented in the limma software package for R, available from the CRAN repository. The methodology is applicable to a variety of microarray experiments and has been shown to be effective in experiments with small sample sizes. The method is particularly useful for experiments with poor quality data or with a limited number of arrays.This paper proposes a method for improving the assessment of differential expression in microarray experiments by utilizing within-array replicate spots. The method estimates the correlation between replicate spots and uses this information to improve the precision of gene-wise variance estimates. This leads to more accurate inference methods for identifying differentially expressed genes. The method is validated using data from a microarray experiment involving calibration and ratio control spots in conjunction with spiked-in RNA. The results show that the common correlation method results in substantially better discrimination of differentially expressed genes from those which are not compared with simply averaging the replicate spots. The method may be combined with empirical Bayes methods for moderating the gene-wise variances between genes. The method is implemented in the limma software package for R, available from the CRAN repository. The methodology is applicable to a variety of microarray experiments and has been shown to be effective in experiments with small sample sizes. The method is particularly useful for experiments with poor quality data or with a limited number of arrays. The common correlation method is shown to have greater power in detecting differential expression while producing no more false positives on average. The method is also shown to be effective in experiments with very few arrays or with poor quality data. The method is implemented in the limma software package for R, available from the CRAN repository. The methodology is applicable to a variety of microarray experiments and has been shown to be effective in experiments with small sample sizes. The method is particularly useful for experiments with poor quality data or with a limited number of arrays.