The accurate modeling of continuum absorption is crucial for the so-called window regions of the spectrum, the relatively transparent regions between lines. The window regions in the microwave are of critical importance for Earth remote sensing and data assimilation. Presented in this paper is an evaluation of the widely used Mlawer, Tobin, Clough, Kneizys, and Davis (MT_CKD) water vapor continuum model in the microwave region, performed using measurements from ground-based radiometers operated by the Department of Energy's Atmospheric Radiation Measurement Program at sites in Oklahoma, USA, and the Black Forest, Germany. The radiometers used were the Radiometrics 23.8/31.4-GHz microwave radiometers (MWRs), the Radiometer Physics GmbH 90/150-GHz MWR at high frequencies (MWRHF), and the Radiometrics 183 GHz G-band vapor radiometer profiler (GVRP). Radiometer measurements were compared with brightness temperatures calculated using radiosonde temperature and humidity profiles input to the monochromatic radiative transfer model (MonoRTM), which uses the MT_CKD continuum model. Measurements at 23.8 GHz were used to correct for biases in the total precipitable water vapor (PWV) from the radiosondes. The long-term 31.4 GHz data set, with a range of PWV values spanning from 0.15 to 5 cm, allowed the separation of uncertainties in the self- and foreign-broadened components of the water vapor continuum. The MT_CKD model has been updated in the microwave region to provide improved agreement with the measurements. MonoRTM has been updated accordingly. The results for the different instruments and frequencies were consistent, providing high confidence in the continuum updates. The estimated uncertainties on the updated continuum coefficients in MT_CKD are 4% on the foreign-broadened water vapor continuum and 4% on the self-broadened water vapor continuum.