One of the causes of public discomfort and complaint about odour in China is the nuisance odour, generated from the municipal sewage treatment plants. With the ability to be dispersed over a long distance, the odours can affect a large number of people. With the aim of identifying the compounds contributing the most to the overall odour emanating from municipal sewage treatment plant, and developing a prediction model for sensory odour concentration based on the compound odour activity value (OAV), odour samples from 2 days were collected at a municipal sewage treatment plant in Tianjin in the months of October and November 2013. Odour concentrations (OCs) were measured by the triangular odour bag method. Chemical components were quantified by gas chromatography–mass spectrometry. According to the analysis of odour emission characteristics, it was found that hydrogen sulfide and methyl mercaptan were the key odorants responsible for the overall odour. To understand the interrelationship of these two odorants, 10 groups of a binary mixture of hydrogen sulfide and methyl mercaptan, representing different levels of odour concentration and intensity, were prepared in the laboratory. OCs were regressed against OAV using multivariate linear regression. A statistically significant positive correlation was found between single-compound OAV and odour concentration (by both SPSS and Minitab software). Furthermore, the models were validated by field monitoring data, which showed the odour prediction concentration had a good fit to the measured concentration by using Minitab software. Lastly, the Austal 2000 model system was used for the simulation of the odour emission dispersion into the surrounding area. This study provides an effective way to predict the odour emission condition in municipal sewage treatment plant.