Background: The Australian plague locust, Chortoicetes terminifera, is among the most promising species to unravel the suites of genes underling the density-dependent shift from shy and cryptic solitarious behaviour to the highly active and aggregating gregarious behaviour that is characteristic of locusts. This is because it lacks many of the major phenotypic changes in colour and morphology that accompany phase change in other locust species. Reverse transcription-quantitative polymerase chain reaction (RT-qPCR) is the most sensitive method available for determining changes in gene expression. However, to accurately monitor the expression of target genes, it is essential to select an appropriate normalization strategy to control for non-specific variation between samples. Here we identify eight potential reference genes and examine their expression stability at different rearing density treatments in neural tissue of the Australian plague locust. Results: Taking advantage of the new orthologous DNA sequences available in locusts, we developed primers for genes encoding 18SrRNA, ribosomal protein L32 (RpL32), armadillo (Arm), actin 5C (Actin), succinate dehydrogenase (SDHa), glyceraldehyde-3P-dehydrogenase (GAPDH), elongation factor 1 alpha (EF1a) and annexin IX (AnnIX). The relative transcription levels of these eight genes were then analyzed in three treatment groups differing in rearing density (isolated, short- and long-term crowded), each made up of five pools of four neural tissue samples from 5th instar nymphs. SDHa and GAPDH, which are both involved in metabolic pathways, were identified as the least stable in expression levels, challenging their usefulness in normalization. Based on calculations performed with the geNorm and NormFinder programs, the best combination of two genes for normalization of gene expression data following crowding in the Australian plague locust was EF1a and Arm.We applied their use to studying a target gene that encodes a Ca2+ bin