Expression microarray technology is the ideal tool for the study of thousands of gene expression patterns simultaneously. However there is a great variability of results attributed to technical and statistical analysis aspects. In this work we present several of the arisen problems of a differential expression study in liver of Iberian pigs under the treatments sex and feeding level using Affymetrix microarray. Normalized expression data were analyzed following two approaches of the mixed model methodology. In both treatments the detected differential expressions were dependent of the statistical model and just a small number of genes were coincident between both statistical strategies. Some of the expression differences were confirmed by quantitative PCR. Besides, we have identified design mistakes and missing annotation of the array probes. The results of this study have allowed us to detect differential expression of interesting genes, but it pointed out the necessity of carrying out complementary studies in order to confirm the differential expressions revealed using microarrays technology.