Red Midcourse Space Experiment (MSX) Sources (RMSs) are regarded as excellent candidates of massive star-forming regions. In order to characterize the chemical properties of massive star formation, we made a systematic study of 87 RMSs in the southern sky, using archival data taken from the Atacama Pathfinder Experiment Telescope Large Area Survey of the Galaxy (ATLASGAL), the Australia Telescope Compact Array, and the Millimetre Astronomy Legacy Team Survey at 90GHz (MALT90). According to previous multiwavelength observations, our sample could be divided into two groups: massive young stellar objects and HII regions. Combined with the MALT90 data, we calculated the column densities of N_2_H^+^, C_2_H, HC_3_N, and HNC and found that they are not much different from previous studies made in other massive star-forming regions. However, their abundances are relatively low compared to infrared dark clouds (IRDCs). The abundances of N_2_H^+^ and HNC in our sample are at least 1mag lower than those found in IRDCs, indicating chemical depletions in the relatively hot gas. Besides, the fractional abundances of N_2_H^+^, C_2_H, and HC_3_N seem to decrease as a function of their Lyman continuum fluxes (N_L_), indicating that these molecules could be destroyed by UV photons when HII regions have formed inside. We also find that the C_2_H abundance decreases faster than HC_3_N with respect to N_L_. The abundance of HNC has a tight correlation with that of N_2_H^+^, indicating that it may be also preferentially formed in cold gas. We regard our RMSs as being in a relatively late evolutionary stage of massive star formation.