The infiltration approach to watershed runoff is acceptable for those situations where rain in excess of infiltration causes surface runoff. It cannot be applied rigorously to urban watersheds where little infiltration takes place, or to highly pervious soils where little excess occurs. A mathematical form is proposed for empirical calibration on real-world heterogeneous watersheds. The effect of variation of parametric values is explored and some optimized values, resulting from fitting to plot infiltration data, are presented.