Twenty-four high and 24 low yielding inbred lines per se were selected from a random group of 247 inbred lines from the 'Iowa Stiff Stalk Synthetic' maize (Zea mays L.) population to produce two groups of lines which differ in gene frequency. High x high (HH), High x low (HL) and low x low (LL) single-cross hybrids were produced to determine the relationship of hybrid performance to heterosis, dominance and gene frequency. The high and low yielding lines were also assayed with 13 enzyme loci to determine if a relationship exists between genetic diversity assessed by the enzyme loci and hybrid performance and heterosis within and among the three groups of hybrids. Theory was developed to relate hybrid performance and heterotic deviations to additive and dominance genetic parameters, which were estimated. There were significant differences among the HH, HL, and LL group means and heterotic deviations for yield and days to silk. Estimates of means and variances suggested dominance for yield is in the partial to complete range and that the average frequency of the favorable allele for yield in BSSS is greater than 0.5. There were no correlations between inbred traits and hybrid yield within groups, suggesting inbred traits are of little value in selecting for hybrid yield. Selection among lines per se was effective at dividing the lines into groups with high and low average performance in hybrids. However, selecting the highest yielding line per se within groups does not identify the line with the highest average performance in hybrids. Two of enzyme loci (Got1 and Enp1) were fixed and two others (Est4 and Pgd2) were nearly fixed leaving 9 enzyme loci, which represented five of the 10 maize chromosomes. The results indicated that enzyme diversity is not related to either hybrid performance or midparent heterosis. Our results and those of others suggest that enzyme loci are of limited usefulness in predicting combining ability in maize.