We have made new models for mass-losing OH/IR stars to explain the properties of the dust shells around them using more accurate information about the material in the shell and the physical precesses including pulsations. We have applied our dust opacity which has been deduced from observations and experimental data to various density distributions, calculated the model emergent spectra, and compared with observations. Contrary to previous suggestions, we could fit observations fairly well using density distribution $ hopropto r^{-2}$, which is physically plausible, with proper choice of opacities. The time scales for dust formation, growth, and movement are calculated to be compared with pulsation periods. The change of the emergent spectrum depending on the phase of pulsation can be explained fairly well by changing dust condensation radius(for fixed dust condensation temperature) in step with the change in stellar luminosity. The effects of stellar wind models and pulsation models on dust shells with attention to emergent spectra are discussed.