The object of this thesis is the synthetic generation of possible future flows in the James River at Buchanan, Virginia, through the utilization of existing records of monthly precipitation and runoff. Twelve separate monthly regression models were chosen on the basis of the degree of correlation as indicated by the correlation coefficient. Random monthly precipitation totals were generated from the twelve separate distributions of historical monthly precipitation totals. The synthetic precipitations were used in the twelve monthly models to generate 5000years of artificial record.

The 5000-year synthetic record was divided into fifty separate 100-year records for the purpose of analysis. The dependable flows were found for each of the fifty records for each of eight reservoir sizes. The ninety percent confidence interval and the dependable flow for each of eight reservoir sizes was found.

The model had a negative bias of approximately thirteen percent. This bias appeared to be alleviated when the dependable flow was expressed as a percent of the average flow for the period in question. It appears that there is great variability in the estimated value of the dependable flow in the James River.