Best estimates of evaporation at Williams Lake, north central Minnesota, were determined by the energy budget method using optimum sensors and optimum placement of sensors. These best estimates are compared with estimates derived from using substitute data to determine the effect of approximately biweekly periods using five open water seasons. For most of the data substitutions that affected the Bowen ratio, new values of evaporation differed little from best estimates. The three data substitution methods that caused the largest deviations from the best evaporation estimates were (1) using changes in the daily average surface water temperature as an indicator of the lake heat storage term, (2) using shortwave radiation, air temperature, and atmospheric vapor pressure data from a site 110 km away, and (3) using an analog surface water temperature probe. Recalculations based on these data substitutions resulted in differences from the best estimates as much as 89%, 21%, and 10%, respectively. The data substitution method that provided evaporation values that most clearly matched the best estimates was measurement of the lake heat storage term at one location in the lake, rather than at 16 locations. Evaporation values resulting from this substitution method usually were within 2% of the best estimates. ¿ American Geophysical Union 1993