This paper first examines the commonly-used thermal-moisture analogy approach in thermal-moisture analogy approach. We conclude that such an analogy using a normalized concentration approach does not exist in the case of soldering reflow, when the solubility of each diffusing material varies with temperature or the saturated moisture concentration is not a constant over an entire range of reflow temperatures. The whole field vapor pressure distribution of a flip chip BGA package at reflow is obtained based on a multiscale vapor pressure model. Results reveal that moisture diffusion and vapor pressure have different distributions and are not proportional. The vapor pressure in the package saturates much faster than the moisture diffusion during reflow. This implies that the vapor pressure reaches the saturated pressure level in an early stage of moisture absorption, even the package is far from moisture saturated. However, the interfacial adhesion degrades continuously with moisture absorption. Therefore, the package moisture sensitivity performance will largely reply on the adhesion strength at elevated temperature with moisture. A specially designed experiment with a selection of six different underfills for flip chip packages was conducted. Results confirm that there is no correlation between moisture absorption and the subsequent interface delamination at reflow. The adhesion at high temperature with moisture is the only key modulator that correlates well with test data. Such a parameter is a comprehensive indicator, which includes the effects of thermal mismatch, vapor pressure, temperature and moisture. In this paper, a micromechanics based mechanism analysis on interfacial delamination is also presented. With the implementation of interface properties into the model study, it shows that the critical stress, which results in the unstable void growth and delamination at interface, is significantly reduced when the effect of moisture on debonding is considered.