Predictive models that estimate seed mass of hydrophytes are useful for efficiently and accurately estimating forage availability for waterfowl in wetlands. Two predictors of seed mass have been successfully used in the past: a dot grid method that places plant inflorescences on a 9 dots/cm² grid and counts the number of dots obscured by seed and measuring multiple plant measurements (e.g., inflorescence length and width). I developed 6 equations to predict seed mass/plant for Walter's millet (Echinochloa walteri) from 9 impoundments on 5 National Wildlife Refuges located in New Jersey, Delaware, and Virginia, USA. These equations included variables for number of dots obscured by seeds, plant morphology measurements, geographic location, and combinations of these variables. I used Akaike's Information Criterion to compare these competing models. The model using the number of dots covered by seeds (Y [g/plant] = 0.0036 × number of dots obscured) performed the best among the models tested. This model explained 85.9% of the variation in seed mass. The dot-grid method is a reliable and efficient technique to estimate seed mass of hydrophytes. I recommend that wetland managers use the dot-grid method to estimate available seed forage for waterfowl and evaluate this technique for additional plant species and geographic areas. (WILDLIFE SOCIETY BULLETIN 34(1):156–158; 2006)