Naegleria gruberi is a free-living amoeboflagellate commonly found in freshwater and in soils around the world. It is a non-pathogenic relative of Naegleria fowleri, which is the etiologic agent of Primary Amoebic Meningoencephalitis (PAM). PAM occurs world-wide and it is considered a rare disease, but its fatality rate is high (96%) mainly because of delay in initiation of treatment due to misdiagnosis and lack of a specific treatment. The analysis of gene expression by quantitative real-time PCR in N. gruberi could be a highly efficient means to understand the pathogenicity of N. fowleri and also to find drug targets. Accurate RT-qPCR analysis requires correct normalization of gene expression data using reference genes (RG), whose expression should be constant under different experimental conditions. In this study, six genes, representing the most frequently used housekeeping genes, were selected for evaluation as reference genes in N. gruberi. The expression and stability of these genes was evaluated employing four algorithms (geNorm, NormFinder, BestKeeper and RefFinder). This work shows significant variations of the stability of RGs depending on the algorithms employed and on the experimental conditions (i.e. logarithmic, stationary, heat-shock and oxidative stress). The geNorm, NormFinder and RefFinder analysis of all the experimental conditions in combination revealed that ACT and G6PD were the most stable RGs. While BestKeeper analysis showed that 18S and TBP were the most stable RGs. Moreover, normalization of HSP90 gene expression with the most stable RGs resulted in an upregulation whereas when the normalization was done with the unstable RGs, the gene expression was not reliable. Hence, the implications of this study are relevant to gene expression studies in N. gruberi.