Development of maize populations resistant to stem borers depends largely on the existence of useful genes or alleles, which can combine to confer resistance to progenies. Such genes are often available in areas of stress, having been responsible for the survival of such crops over the years. Pink stem borer, Sesamia calamistis (Hampson, Noctuidae) and sugarcane borer, Eldana saccharina (Walker, Pyralidae) are endemic in southeastern Nigeria. Damages caused by the larvae of these moths are more prevalent during the second planting season (August-November). Genetic diversity for a range of agronomic and resistance attributes within 209 local maize collections from southeastern Nigeria and 3 improved check varieties were investigated in field trials in randomised complete block design (RCBD) with two replications across three environments. Data collected from the evaluations were subjected to both uni- and multivariate statistics. Furthermore, four traits namely, leaf feeding, ear damage, shoot breakage and yield were used from across three environments to construct a selection index. The multivariate analysis on the plant attributes, using canonical discriminant analysis, revealed the agronomic and borer damage parameters that contributed significantly to the total variation observed in different environments. Out of the four canonical discriminant functions obtained, two had significant (P=0.05) eigenvalues accounting for over 98 % of the total variation. The first canonical function was mainly associated with yield while the second was associated with the borer damage attributes. Rank summation index (RSI) used to rank the entries for resistance to stem borers identified 11 genotypes representing top 5 % of the total as resistant. In the second experiment the 11 genotypes and their hybrids, made in a diallel fashion were evaluated for agronomic and borer damage attributes in seven environments in RCBD with three replications. Data collected were subjected to analysis of variance and those found significant (P=0.05) were further subjected to diallel analysis using Griffing’s method 2 model 1 for fixed effects. Significant GCA and SCA effects were obtained for most of the traits studied in the various environments and in the pooled environment thus indicating that additive and non-additive gene effects were involved in the expressions of the traits studied. However, in a few cases, only GCA or SCA was important thus indicating the relative importance of the genetic component of the variance. The assessment of the agronomic and borer damage attributes of the parents and the crosses indicate that the variety crosses were not superior to the parents in most of the traits. The significant differences observed between the parents and the crosses for dead heart and leaf feeding damage parameters is suggestive of the occurrence of exploitable heterosis for the development of genotypes that are resistant to stem borer attack. Genotypes SE NG-33, SE NG-65 and TZBR Syn W had high negative GCA values for dead heart while SE NG-62, SE NG-148, TZBR Syn W and TZBR ELD 3 C2 had the high negative GCA values for leaf feeding damage. For ear damage, SE NG-65, SE NG-67, SE NG-119, SE NG-148 and AMA TZBR-W-C1 had high negative GCA estimates. Genotypes SE NG-33, SE NG-62, SE NG-65, SE NG-77, SE NG-106 and SE NG-119 had the highest positive GCA effects for grain yield. The nine genotypes selected formed two heterotic pools: Group A comprised SE NG-33, SE NG-77, SE NG-106, SE NG-148 and TZBR Syn W while Group B included SE NG-62, SE NG-119, AMA TZBR-W-C1 and TZBR ELD 3 C2. Average yield of the grouped genotypes crossed in all possible combinations was 1.06 t ha-1 showing 5 % yield increase. Furthermore, the best five yielding crosses namely; SE NG-33 x TZBR ELD 3 C2, SE NG-62 x SE NG-77, SE NG-62 x SE NG-106, SE NG-106 x TZBR ELD 3 C2 and TZBR Syn W x TZBR ELD 3 C2, selected may be used as population crosses or in the formation of composite varieties.