Testing of microsatellite instability is not only used as a triage for possible Lynch syndrome, but also to predict immunotherapy treatment response. The aim of this study was to assess the frequency of mismatch repair deficiency (MMR-D)/microsatellite instability (MSI) in 400 cases of non-endometrioid ovarian tumors (high-grade serous, low-grade serous, mucinous and clear cell), to compare different methodological approaches of testing, and to assess the optimal approach for next generation sequencing (NGS) MSI testing. For all tumors, we evaluated immunohistochemical (IHC) expression of MMR proteins and assessed microsatellite markers by PCR-based method. Except for high-grade serous carcinoma, we correlated the findings of IHC and PCR with NGS-based MSI testing. We compared the results with somatic and germline mutation in MMR genes. Among the whole cohort, seven MMR-D cases, all clear cell carcinomas (CCC), were found. On PCR analysis, 6 cases were MSI-high and one was MSS. In all cases, mutation of an MMR gene was found; in 2 cases, the mutation was germline (Lynch syndrome). An additional 5 cases with a mutation in MMR gene(s) with MSS status and without MMR-D were identified. We further utilized sequence capture NGS for MSI testing. Employing 53 microsatellite loci provided high sensitivity and specificity. Our study shows that MSI occurs in 7% of CCC while it is rare or absent in other nonendometrioid ovarian neoplasms. Lynch syndrome was present in 2% of patients with CCC. However, some cases with MSH6 mutation can evade all testing methods, including IHC, PCR, and NGS-MSI.