The correlation between ambient temperature and equivalent real-world fuel economy (FE) for plug-in hybrid electric vehicles (PHEVs) is the subject of an investigation seeking to inform economic, environmental, and policy assessments of these vehicles and technologies. This paper collected seven months of 1-Hz vehicle operating data from 119 Odyne medium-duty PHEV work trucks in commercial service across the United States. The data used for this paper were collected from the vehicle's controller area networks using mobile data collection and transmission devices. The raw data were aggregated into a central database for analysis. We present the methods of data processing, filtering, intermediate data manipulation, and analysis implemented using MATLAB, R, SQL, and Java-based Hadoop MapReduce. Regression analysis is used to calculate models describing the sensitivity of PHEV FE to ambient temperature, and the models are found to be similar to regressions for conventional vehicles (CVs) from other studies. This paper provides evidence that the sensitivity of medium-duty PHEV FE is similar to that of CVs. These results can provide guidance to researchers seeking to model real-world PHEV energy consumption.