Abstract: Trials were conducted at two different locations in the Forest and Transition ecozones of Ghana in 2004 and 2005. The aim was to evaluate the agronomic performance of four elite varieties against four released varieties and to determine GXE effect on selection of varieties for the Forest and Transition ecozones of Ghana. Eight genotypes involving four local elite and four released varieties were used. Data collected include tuber yield, harvest index, dry matter content, cooking quality, flour, starch and gari yields. Data collected were analysed with Analysis of Variance (ANOVA) including the factors genotype, location and year for all the data collected except cooking quality using Costat. The difference method in which experienced but few people are used, was use to rank the cooking quality of the genotypes on a scale of 1 to 4 with 1 as poor and 4 as excellent. Tuber yield was generally higher for the released varieties than the local elite ones. The opposite was true for the dry matter content. Harvest index was in the range of 0.48-0.64. Cooking quality of the local elite varieties was generally better than the released ones The results show differences in response of the agronomic traits of cassava to different environmental conditions. This justifies specific adaptation as a goal for local breeding. It also shows that these traits are genotype dependent. It is therefore, critical that cassava genotypes are screen across the various cassava growing areas to assess their potential usage before its potential for specific use may be recommended.