This paper summarizes a long-term (1963)(1964)(1965)(1966)(1967) study of fallout on a Utah dairy farm. Mathematical models and statistical methods (including non-linear least-squares techniques) were used to obtain estimates of key parameten in a compartmental model. Estimates of the daily rate of change in fallout input were obtained from statistical analysis of fallout collector data and (independently) by fitting an abbreviated compartmental model to observations of 137Cs concentration in five alfalfa harvests. This and other cross-checks indicate internal consistency in model and data. Parameters estimated from the data, and from other sources, were used in a computer model that appears to generate data points very close to those actually observed (standard deviation of 55 pCi/m2 on an individual observation). In later years of the study, the effect of an (assumed) remobilization of 137Cs from alfalfa rootstocks reduces closeness of fit of models and data. In common with other studies, we found interception of fallout to vary with plant yield, suggesting an effect of growth on interception. Results of the Utah study were contrasted with three, somewhat similar investigations (in England, Colorado and Michigan) and while certain differences limit the comparisons, it can be concluded that the overall framework of radionuclide transfer through a dairy system is now reasonably well established.