This article discusses normalization methods for cDNA microarray data. The goal of normalization is to adjust for effects caused by technical variations rather than biological differences between RNA samples or printed probes. The authors describe several normalization techniques, including print-tip loess normalization, which is a well-tested method that adjusts for intensity and spatial trends. This method can be refined by using quality weights for individual spots. Diagnostic plots are recommended to visualize these trends and identify any remaining biases in the data. If biases persist, further normalization steps such as plate-order normalization or scale normalization between arrays may be necessary. Composite normalization is used when control spots are available that are known to be not differentially expressed. Variations of loess normalization include global loess normalization and 2D normalization. The authors also discuss other trends that can be adjusted for in normalization, such as print-order effects and differences in DNA purity between amplification batches or clone libraries. The article provides detailed commands for implementing these normalization techniques using freely available software. It emphasizes the importance of using spot quality weights to improve the reliability of the normalization process. The authors conclude that print-tip loess normalization is a well-tested general-purpose method that works well on a wide range of arrays. The method can be refined by using quality weights and is best combined with diagnostic plots to ensure accurate results.This article discusses normalization methods for cDNA microarray data. The goal of normalization is to adjust for effects caused by technical variations rather than biological differences between RNA samples or printed probes. The authors describe several normalization techniques, including print-tip loess normalization, which is a well-tested method that adjusts for intensity and spatial trends. This method can be refined by using quality weights for individual spots. Diagnostic plots are recommended to visualize these trends and identify any remaining biases in the data. If biases persist, further normalization steps such as plate-order normalization or scale normalization between arrays may be necessary. Composite normalization is used when control spots are available that are known to be not differentially expressed. Variations of loess normalization include global loess normalization and 2D normalization. The authors also discuss other trends that can be adjusted for in normalization, such as print-order effects and differences in DNA purity between amplification batches or clone libraries. The article provides detailed commands for implementing these normalization techniques using freely available software. It emphasizes the importance of using spot quality weights to improve the reliability of the normalization process. The authors conclude that print-tip loess normalization is a well-tested general-purpose method that works well on a wide range of arrays. The method can be refined by using quality weights and is best combined with diagnostic plots to ensure accurate results.