A meta-analysis approach to prediction of farm level yield risk from county level yield series is applied to Kansas wheat yields. A nonlinear relationship between county level and farm level yield risk is found, which indicates that yield risk increases at an increasing rate as the number of acres in the risk measure decreases. County level yield variability should be adjusted upward by approximately .1% for each percent difference in county acreage and average farm acreage within the county. The meta-analysis approach is shown to be promising for the prediction of farm level yield risk when farm level information is difficult to obtain.