We assessed the genetic diversity in Trichloris crinita (Poaceae) varieties from South America, using AFLPs, morphological characters, and quantitative agronomic traits. Owing to the importance of this species for range grazing, we first characterized the varieties based on forage productivity. Biomass production varied 9 fold among the materials evaluated. Analysis of AFLP fingerprints allowed the discrimination of all varieties with a few selected primer combinations. Pair-wise genetic similarities, using marker data, ranged from 0.31 to 0.92 (Jaccard coefficients). Marker-based unweighted pair group method with arithmetic averaging (UPGMA) cluster analysis did not show geographical clustering, but rather grouped the varieties according to their biomass production. We identified 18 markers associated with biomass production, of which 8 showed complete correlation (r = 1.00) with this trait. These DNA markers can be used to assist selection for high forage productivity in T. crinita. Cluster analysis using morphological and quantitative characters revealed 4 distinct groups of varieties, clearly separated according to their biomass yield. The variables foliage height and basal diameter were strongly correlated with biomass production and these phenotypic markers can be used to select productive plants. The relations among the varieties based on AFLP data were significantly correlated with those based on agronomic and morphological characters, suggesting that the 2 systems give similar estimates of genetic relations among the varieties.