Parametric studies using 26 site-years of meteorological data from 10 different nuclear reactor sites were made to determine the sensitivity of diffusion estimates to selected aspects of meteorological measurement programs. These sensitivities were determined by comparing diffusion estimates obtained using the actual data records with estimates obtained by altering the same record to simulate incomplete data recovery and instrument error. Variations from one year to the next were also explored, as was the effect of varying assumed wind speeds for calm hours. Results indicate that sufficiently accurate estimates, well within the accuracy and conservatism of the diffusion models, can be made with a smaller data base and with less measurement accuracy than required by U.S. Atomic Energy Commission guidelines.