We compare mass-loss rates of OH/IR stars obtained from radio observations with those derived from the dust radiative transfer models and IR observations. We collect radio observational data of OH maser and CO line emission sources for a sample of 1533 OH/IR stars listed in Suh & Kwon (2011). For 1259 OH maser, 76 CO(J=1-0), and 55 CO(J=2-1) emission sources, we compile data of the expansion velocity and mass-loss rate. We use a dust radiative transfer model for the dust shell to calculate the mass-loss rate as well as the IR color indices. The observed mass-loss rates are in the range predicted by the theoretical dust shell models corresponding to M = 10-8M⊙/yr - 10-4M⊙/yr. We find that the dust model using a simple mixture of amorphous silicate and amorphous Al2O3 (20% by mass) grains can explain the observations fairly well. The results indicate that the dust radiative transfer models for IR observations generally agree with the radio observations. For high mass-loss rate OH/IR stars, the mass-loss rates obtained from radio observations are underestimated compared to the mass-loss rates derived from the dust shell models. This could be because photon momentum transfer to the gas shell is not possible for the physical condition of high mass-loss rates. Alternative explanations could be the e®ects of di®erent dust-to-gas ratios and/or a superwind.

We investigate the properties of OH, SiO, and H2O maser emission in O-rich AGB stars. We use a sample of 3373 objects, which is an updated version of the list of O-rich AGB stars presented in Suh & Kwon (2011). We divide the 3373 O-rich AGB stars into four different groups based on the maser emission: OH maser sources (1533), SiO sources (1627), H2O sources (452), and sources with no maser (610). To understand the nature of the maser sources, we present various infrared two-color diagrams (2CDs) using IRAS, 2MASS, and AKARI data. For each group, we compare the positions on various infrared 2CDs with theoretical models. We ¯nd that the OH maser sources generally show larger color indices and larger dust optical depths than SiO or H2O sources. We suggest that the differences of the color indices for different maser sources are due to different mass-loss rates and dust formation processes.