The severe combined immunodeficiencies (SCIDs) are a set of life threatening genetic diseases in which patients are born with mutations in single genes and are unable to develop functional immune systems. While allogeneic bone marrow transplantation can be curative for these diseases, there remain significant limitations to this approach. Gene therapy using viral vectors containing a corrective transgene is being developed for some of these disorders; however, for other SCID disorders, such as those caused by genetic mutations in RAG1 and RAG2, the transgene needs to be expressed in a precise, developmental and lineage specific manner to achieve functional gene correction and to avoid the risks of cellular transformation. In contrast to using viral vectors to deliver transgenes in an uncontrolled fashion, we are working towards developing CRISPR genome editing to correct the RAGs disease-causing mutations by precisely modifying the genome. CRISPR genome editing requires delivery of both the Cas9 nuclease and the targeting guide RNA (gRNA). The gRNA component can be generated in multiple ways, each with advantages and disadvantages. Here we compare the efficiency of editing, the on- and off-target repair profiles of RAG1 and RAG2 gRNAs delivered as an in-vitro transcribed (IVT) single guide RNA (sgRNA), a chemically-synthesized sgRNA and a chemically-synthesized bipartite complex (crRNA + tracrRNA). Our results show that the chemically-modified sgRNAs and the bipartite crRNA + tracrRNA complex, delivered as a ribonucleoprotein (RNP) complex, enable the highest genome editing in human primary CD34+ hematopoietic stem and progenitor cells (HSPCs) with lowest toxicity. Additionally, we show that we can use the combination of CRISPR-Cas9 RNP, chemically modified gRNAs, and recombinant adeno-associated viral vector (rAAV) donor transduction to effectively target functional RAG2 cDNA into the endogenous locus in human primary CD34+HSPCs. We will also present a summary of a comprehensive analysis of the off-target events associated with the delivery of the synthetic RAG1 and RAG2 gRNA forms. The off-target profiles for each class of gRNA will be compared using the unbiased GUIDE-seq approach and quantified using rhAmpSeq, a multiplexed, amplification-based, target enrichment next-generation sequencing (NGS) approach and finally the implication of our findings for therapeutic genome editing will be discussed.