Abstract：In view of the limitations of the currently widely used method for the cDNA microarray data standardization (often called log-ratio transformation method), a new method was proposed which named as non-transformation method. It does not transform the data but directly estimates the gene transcription level after correction of the background effect based on Huber's model and principle of least square estimation through an iteration process. Using computer simulation, the new method was compared with the log-ratio method under various conditions (different number of repeated probes on a chip, different background variation, and different ratio of up- and dawn-regulated expressed genes). The results showed that in all conditions the non-transformation method performed better than the log-ratio method. The relative biases from the non-transformation method were less than 5%, while those from the log-ratio method more than 20%. Furthermore, the log-ratio method was very sensitive to the background variation and the ratio of up- and dawn-regulated expressed genes. With the increase of the background variation or the unbalance level of the ratio of up- and dawn-regulated expressed genes, the biases from the log-ratio method increased rapidly, while the non-transformation method was almost not influenced by these factors. Therefore, the non-transformation method could be used as an alternative method for cDNA microarray data standardization
