Although reference genes have previously been used in the expression analysis of genes involved in sugarcane flowering they had not been experimentally validated for stability and consistency of expression between different samples over a wide range of experimental conditions. Here we report the analysis of candidate reference genes in different tissue types, at different temporal time-points, in both short and long day photoperiodic treatments. The stability of the candidate reference genes in all conditions was evaluated with NormFinder, BestKeeper, and RefFinder algorithms that complement each other for a more robust analysis. As the Normfinder algorithm was more appropriate for our experimental conditions, greater emphasis was placed on Normfinder when choosing the most stable genes. UBQ1 and TUB were shown to be the most stable reference genes to use for normalizing RT-qPCR gene expression data during floral induction, whilst 25SrRNA1 and GAPDH were the least stable. Their use as a reference gene pair was validated by analyzing the expression of two differentially expressed target genes (PIL5 and LHP1). The UBQ1/TUB reference genes combination was able to reveal small significant differences in gene expression of the two target genes that were not detectable when using the least stable reference gene combination. These results can be used to inform the choice of reference genes to use in the study of the sugarcane floral induction pathway. Our work also demonstrates that both PIL5 and LHP1 are significantly up-regulated in the initial stages of photoperiodic induction of flowering in sugarcane.