Twenty-four high- and 24 low-yielding lines per se selected from 247 random inbred lines of the ‘Iowa Stiff Stalk Synthetic‘ (BSSS) maize (Zea mays L.) population were used to produce 48 high ✕ high (HH), 96 high ✕ low (HL), and 48 low ✕ low (LL) single-cross hybrids. The objectives of the study were to determine how crosses of lines selected for yield per se perform, and to examine the relationship of per se and hybrid performance. The 192 hybrids were evaluated in two replications at each of six Iowa environments. There were significant differences among HH, HL, and LL hybrid group means for grain yield and they were ranked (HH > HL > LL) as expected under a model assuming partial to complete dominance. The data suggests that on the average overdominance was not important for yield in BSSS. Correlations of inbred line traits with the same traits of the hybrids were larger than correlations of inbred line traits with hybrid yield. The correlations of inbred traits with hybrid yield were too small to be of predictive value. The data suggested that selection among lines per se for yield could be used to separate a population into a group of high and low combining lines based on their per se performance. However, per se performance within groups was not related to either specific or general combining ability. After selection for per se performance the smaller group of selected lines could be crossed to a series of testers to identify lines with the greatest hybrid potential.