By modeling synthetic Pn amplitudes, I and my colleagues in 2007 proposed a Pn geometric-spreading model (Y2007) that takes into account the spherical shape of the Earth. In this study, I used a set of observed Pn amplitudes from the tectonically active regions of Asia to evaluate the performance of Y2007 and to develop new, observation-based Pn spreading models. Even though Y2007 provides improved geometric-spreading correction of Pn amplitudes over the traditional power-law model, the corrected amplitudes exhibit undesirable decay rate variations. To address this issue, I used a procedure to develop Pn spreading models based on observed data. I first correct the Pn amplitudes for attenuation using an average quality factor Q estimated from Y2007-corrected Pn amplitudes. I then develop a spreading model, which is a simplified version of Y2007, by fitting the corrected amplitudes. Compared with Y2007, the new spreading model significantly reduces amplitude variations, particularly at short distances. To more accurately model the complex data behavior, I also developed a segmented spreading model in which separate sets of model parameters are derived for amplitudes in different distance ranges. The spreading models developed in this study account for radially symmetric elastic and other effects, such as velocity gradient, forwarding scattering, and potential depth-dependent attenuation variation, as well as wavefront expansion and the spherical shape of the Earth. Using the new model for spreading correction results in better attenuation isolation and allows amplitudes in a broader distance range to be used in the accurate mapping of lateral attenuation variations. The method I employed in this study could be used as a general procedure to develop observation-based Pn spreading models for other regions.