ABSTRACT. The efficiency of cultivar trial networks is an important subject in official cultivar testing. We investigated this efficiency for malting barley (Hordeum vulgare L.) in Uruguay, using data on 213 cultivars tested across an eight-year period at six locations. The variance-components approach was used to quantify the effects of years, locations, sowing dates and replicates on the precision of cultivar mean comparisons. The relationships among testing environments and genotypic adaptation patterns were explored via biplots. Factorial regression was used to model genotype × environment interaction (GEI) directly in relation to measured environmental variables. Variance components indicated that both the number of locations and sowing dates could be reduced. Biplot analysis identified some repeatable GEI patterns. Factorial regression showed that mean daily temperature during the emergence-heading period and daily minimum temperature at heading explained 20% of GEI. Still, the majority of the GEI appeared to be highly nonrepeatable. A future network should focus on wide adaptation while enhancing the chances to exploit specific adaptation to the prevalent temperature conditions by sampling contrasting sowing dates at different locations. © 2008. Crop Science Society of America