The generation of complex tissue organoids via directed differentiation of human pluripotent stem cells brings the prospect of personalised drug testing, disease modelling and regenerative medicine. We have developed a protocol for the generation of kidney organoids comprised of nephrons, collecting duct, vasculature and surrounding interstitium (Takasato et al, Nature, 2015) from human pluripotent stem cells. This protocol relies upon the stepwise recapitulation of morphogenetic events previously characterised during normal kidney development in the mouse. The utility of the protocol for applications such as the modelling of human kidney disease will rely implicitly on the accuracy of this differentiation program, the reproducibility of the directed differentiation, the transferability between iPSC lines and the functional authenticity of the human cell types generated. Using RNA-seq based transcriptional analyses across the differentiation timecourse from pluripotent iPSC to kidney organoid, we have examined transcriptional variation between individual organoids, distinct differentiation experiments and between cell lines. Gene ontology interrogation of unsupervised clusters of synexpression, as well as pairwise temporal changes in gene expression, show a clear transition through appropriate intermediate developmental patterning events, including primitive streak, intermediate mesoderm, metanephric commitment and nephrogenesis. This confirms the high correlation between human and mouse metanephrogenesis, and provides a framework for quality control. A transcriptional correlation of >95% was observed between organoids and >90% between distinct differentiation experiments. Distinct iPSC clones also conformed to a matching transcriptional program. An examination of the most variable genes shows that these represent genes characteristic of nephron patterning, suggesting differential organoid maturation as the major source of both inter-experimental and intra-clonal variation. This data provides a framework for removing sources of unwanted experimental variation in the analysis of expression data, thereby increasing the utility of this approach for personalised medicine and functional genomics.