Real-time PCR has become a powerful tool for the detection of inflammatory parameters, including cytokines. Reference or housekeeping genes are used for the normalization of real-time RT-PCR results. In order to obtain reliable results, the stability of these housekeeping genes needs to be determined. In this study the stability of five genes, including beta-actin, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), hypoxanthine phophoribosyl-transferase (HPRT), ubiquitin (UB) and glucose-6-phosphate dehydrogenase (G6PDH), was determined in a lipopolysaccharide inflammation model in chickens. beta-Actin appeared to be the most stable single gene in our model. Because the use of a single gene for normalization can lead to relatively large errors, the use of the geometric mean of multiple reference genes or normalization factor is preferred. The most stable combination for gene expression analysis in this lipopolysaccharide inflammation model in chickens is G6PDH and UB, since their correlation coefficients were 0.953 and 0.969, respectively (BestKeeper) and an M value of 0.34 and a low V(2/3) value of 0.155 (geNorm) were obtained. The use of HPRT and GAPDH should be avoided. The stable housekeeping genes, G6PDH and UB together, can be used to normalize the expression of pro-inflammatory cytokines in a lipopolysaccharide inflammation model in chickens.