This study determined the efficiency of using spaced plants to indirectly improve sward yield and nutritional quality in tall fescue. Narrow-sense heritabilities, genetic and rank correlations, and selection efficiencies were estimated from spaced-plant and seeded-sward evaluation of a tall fescue (Festuca arundinacea Schreb.) breeding population grown at Logan, UT. Heritability for forage yield was similar between spaced plants and swards (0.43 and 0.44, respectively), but genetic correlation between the two was low (0.41). A lack of relationship (r = 0.30; NS) in family ranking between the two spacings further suggested that spaced plants were not accurately predicting sward yield. There was moderate agreement between the spacings for most fiber traits, but a large discrepancy in heritability and low correlation for crude protein. Direct selection using swards was always more efficient than indirect selection of spaced-plants. This was further complicated by relatively low to moderate rank correlations between spaced-plants and swards. Overall, we concluded that spaced plant evaluation appears to be less effective at improving sward yield, and only moderately predictive of sward nutritional quality in tall fescue. New techniques are needed that maximize genetic expression, but simulate actual sward production of forage grasses.