In climate reconstructions by data assimilation, the sensitivities to both proxies and prior estimates need to be taken into account because models are uncertain and proxies are limited spatiotemporally. This study examines these sensitivities using multiple climate model simulations and different combinations of proxies (corals, ice cores, and tree-ring cellulose). Experiments were conducted based on an offline data assimilation approach. These experiments show annual variations in the global distribution of surface air temperature and precipitation range from 850 to 2000. The results indicate that standard deviations of surface air temperature and precipitation amount during the entire period differ by up to 50% due to prior estimates. Experiments with different types of proxies show that the El Niño-like distribution of positive anomalies in the central to eastern tropical Pacific can be reproduced adequately in experiments with corals, but not in experiments without corals. The correlation coefficient of the NINO.3 index from 1971 to 2000 between experiments with corals and the Japanese 55-year Reanalysis (JRA-55) were 0.79 at maximum, while the correlation coefficient between experiments without corals and JRA-55 were 0.20 at maximum.