The recent emerge of site-directed genome editing technologies such as CRISPR-Cas9 is of great interest for research in medicine. In multiple organisms, it has been shown extensively that the technology can be used to generate knock-out disease models with high efficiencies, enabling the study of gene function. However, to generate disease-relevant models, the introduction of specific base pair alterations (knock-in) is desired, since numerous diseases are caused by specific point mutations, leading to amino acid substitutions or splicing defects. Also in zebrafish, CRISPR-Cas9 has been successfully applied to generate knock-out models. However, results from a limited number of small scale studies, showed that the efficiency of CRISPR-Cas9 mediated knock-in approaches to introduce point mutations, is relatively low. These approaches are based on CRISPR-Cas9 stimulated homology-directed repair (HDR), using either single-stranded oligodeoxynucleotide (ssODN), double stranded DNA (dsDNA) or plasmid templates with varying lengths of homology arms. Because there is no clear consensus concerning the most optimal approach, we assessed the suitability of different types of donor templates (ssODNs, dsDNA and plasmids) to introduce specific point mutations at multiple sgRNA target sites in the zebrafish genome by means of CRISPR-Cas9 stimulated homology-directed repair (HDR). Furthermore, we evaluated the influence of homology arm length of the different templates on mutagenesis efficiency and we tested if there is a difference in mutagenesis efficiency between sense and antisense ssODNs at each of the target sites. Finally, we assessed the influence of several chemical compounds, which either block the non-homologous end joining (NHEJ) pathway or stimulate the HDR pathway, on HDR-based knock-in efficiency.