Presence of farmer preferred traits and understanding their sensitivity to heterogeneous environments is important and critical to variety adoption. The study aimed at determining genotype x environment interaction (GEI) effect and heritability of farmers preferred cassava traits. A 5 x 2 α-lattice design with two replicates was used. Ten cassava varieties were evaluated at Kenya Agricultural Research Institute farms in Kakamega and Alupe and on-farm in Mumias, Busia and Teso between 2009 and 2011. Data were collected on; plant height (PT), height of first branching (HB), internode length (IL), Branching level (BL) per plot, total number of storage roots per plant (NR) and Fresh storage roots yield (RY) converted to yield per hectare. Four stability analyses parameters; Finlay and Wilkinson’s regression coefficient (β1), mean squares deviations from the regression (Sdi), Wricke’s ecovalence (Wi) and Additive Main effects and Multiplicative Interaction stability value (ASV) were used. Heritability was estimated as the proportion of phenotypic variances that is due to genetic differences among varieties. GEI effects were significant (p=0.05) for all traits evaluated except PT and NR. Traits related to plant architecture had fairly high heritability as compared to RY and NR with heritability of 6.9% and 13.6%, respectively. W i, Sdi , and ASV significantly (p=0.05) correlated for all Original Research Article American Journal of Experimental Agriculture, 4(6): 686-702, 2014 687 the traits evaluated. However, βi did not significantly (p=0.05) correlate with all the other procedures except with Sdi for height of first branching with Spearman’s rank correlation coefficient of 0.74. For effective cassava evaluation, traits related to plant type which has high heritability should be evaluated early when planting materials is still a hindrance to carry out multi-location replicated trials. Later, with enough planting materials, evaluation should focus on yield and yield related traits with low heritability in multi-location replicated trials. The presence of GEI implies cassava variety selection should be based on both mean performance and stability. Any one of the Wi, Sdi or ASV stability analysis methods can be used.