AbstractAn atmospheric regional climate model (CCLM) was employed to dynamically downscale atmospheric reanalyses (NCEP/NCAR 1, ERA 40) over the western North Pacific and South East Asia. This approach is used for the first time to reconstruct a tropical cyclone climatology, which extends beyond the satellite era and serves as an alternative data set for inhomogeneous observation-derived records (Best Track Data sets). The simulated TC climatology skillfully reproduces observations of the recent decades (1978–2010), including spatial patterns, frequency, lifetime, trends, variability on interannual and decadal time scales and their association with the large-scale circulation patterns. These skills, facilitated here with the spectral nudging method, seem to be a prerequisite to understand the factors determining spatio-temporal variability of TC activity over the western North Pacific. Long-term trends (1948–2011 and 1959–2001) in both simulations show a strong increase of intense tropical cyclone activity. This contrasts with pronounced multidecadal variations found in observations. The discrepancy may partly originate from temporal inhomogeneities in atmospheric reanalyses and Best Track Data, which affect both the model-based and observational-based trends. An adjustment, which removes the simulated upward trend, reduces the apparent discrepancy. Ultimately, our observational and modeling analysis suggests an important contribution of multi-decadal fluctuations in the TC activity during the last six decades. Nevertheless, due to the uncertainties associated with the inconsistencies and quality changes of those data sets, we call for special caution when reconstructing long-term TC statistics either from atmospheric reanalyses or Best Track Data.