This paper presents a validation of indoor path loss models as well as a characterization of large-scale fading for an office propagation topology, based on measurements conducted at 3.5 GHz. The intrinsic channel characteristics and the impact of shadow fading on the reliability of signal prediction were taken into consideration. The mean relative error was calculated to prove the limitations of area-mean, distance-dependent path loss modeling. The distribution of the local mean values of the received power was investigated. Losses from walls and other objects were measured and compared to respective values for the indoor 2.4 GHz channel. Results confirm the impact of obstacle losses on the reliable estimation of shadow depth, as well as their dependence on frequency.