Long term climate data are vitally important in reliably assessing water resources and water related hazards, but in-situ observations are generally sparse in space and limited in time. Although there are several global datasets available as substitutes, there is a lack of comparative studies about their suitability in different parts of the world. In this study, to find out the reliable century-long climate dataset in South Korea, we first evaluate multi-decadal reanalyses (ERA-20 cm, ERA-20c, ERA-40 and NOAA 20th century reanalysis (20CR)) and gridded observations (CRUv3.23 and GPCCv7) for monthly mean precipitation and temperature. In the temporal and statistical comparisons, CRUv3.23 and GPCCv7 for precipitation and ERA-40 for temperature perform the best, and ERA-20c and 20CR also indicate meaningful agreements. For ERA-20 cm, it has only a statistical agreement, but the mean has the difficulty in representing its ensemble. This paper also shows that the applicability of each dataset may vary by region and all products should be locally adjusted before being applied in climate impact assessments. These findings not only help to fill in the knowledge gaps about these datasets in South Korea but also provide a useful guideline for the applicability of the global datasets in different parts of the world.