Widespread use of legume green manures is limited by a lack of reliable varieties that meet the needs of farmers. Improving symbiotic N fixation (SNF) in open-pollinated legume species poses a challenge because normal methods used for assessing SNF conflict with breeding practices such as seed production and removing low-quality sires. We investigated methods for measuring symbiotic N fixation that are compatible with breeding practice for two major legume cover crops: hairy vetch (Vicia villosa) and crimson clover (Trifolium incarnatum L.). We collected two non-destructive samples at the early-flowering stage (Meristem and Wedge) and a sample at seed harvest (Seed). Our objective was to determine which combination of measurements best predicted symbiotic N fixation (%Ndfa) and total N content at 50% flowering. Our statistical approach accommodated the problem that target traits and seed measurements cannot be measured in the same individuals. Overall, the Wedge sample provided the best prediction for Ndfa at 50% flowering. The best predictions for total N plant-1 came from combining vigor ratings with aboveground biomass at seed harvest. In addition to specific recommendations for legume cover crop breeding programs, we provide a flexible statistical method for studying relationships between traits that cannot be measured jointly.