The usage of heritage buildings must match tight requirements about indoor thermal and hygrometric conditions for preservation of the valuable structures. The design of a suitable air conditioning system cannot be done without a deep knowledge of the actual state. The paper shows a new method to achieve the needed data more effectively. The case study is the condition monitoring of a library at the first floor of the main tower of the Masino Castle (TO). A set of ring-shaped decorated rooms, hosted the library since the eighteenth century. However, poor indoor conditions forced the owner of the manor (FAI) to transfer the books to a more suitable environment. The standard microclimatic monitoring conducted after the restorations shows a relative humidity (RH) always above 65%, threshold suggested by International Federation of Library Associations and Institutions (IFLA) for the preservation of library materials. Therefore, a research aimed at preventing degradation through the evaluation and control of microclimatic conditions has been running. The project proposes a global view, instead of single measurements of temperature and RH. A thermographic technique, underlying a thermodynamic model, allows detecting the wall temperatures and also air velocity, air temperature and RH. The density of data given by this technique is much higher than possible with traditional sensors. Moreover, this innovative method becomes an essential tool for mapping the internal air and humidity flows and especially their development, often due to degradation, which are instead mistakenly attributed to factors endogenous to the structures. The measurements allow quantifying effects of airflow and influence of the openings. The use of a computational fluid dynamic code allows also predicting effects due to the users and visitors on the microclimate. Finally, results are presented in a three-dimensional corrected geometry of the building. Such a data allows a much better design of the air conditioning system, based on specific volumetric data.