This study compares eight background correction methods for two-color microarrays to identify the best option for differential expression analyses. The methods are evaluated based on precision, bias, and their ability to detect differentially expressed genes using popular algorithms SAM and limma eBayes. The traditional local background subtraction method is found to produce negative corrected intensities and high variability of low-intensity log-ratios. New methods, such as normexp+offset, morph, and vsn, which stabilize the variance of log-ratios, perform better. Normexp+offset is the most effective method, followed by morph and vsn. These methods produce strictly positive corrected intensities and avoid missing values, making them superior to the standard method. The study also highlights the importance of balancing bias and precision in background correction to achieve optimal differential expression detection.This study compares eight background correction methods for two-color microarrays to identify the best option for differential expression analyses. The methods are evaluated based on precision, bias, and their ability to detect differentially expressed genes using popular algorithms SAM and limma eBayes. The traditional local background subtraction method is found to produce negative corrected intensities and high variability of low-intensity log-ratios. New methods, such as normexp+offset, morph, and vsn, which stabilize the variance of log-ratios, perform better. Normexp+offset is the most effective method, followed by morph and vsn. These methods produce strictly positive corrected intensities and avoid missing values, making them superior to the standard method. The study also highlights the importance of balancing bias and precision in background correction to achieve optimal differential expression detection.