The analysis of lines S-l and S-2 and the regression of the measurements of the S-2 on their corresponding S-l were used to estimate the existing genetic variability in a Nebraska Stiff Stalk Synthetic (NSS) corn population at two localities, Mead and Lincoln, Nebraska-USA. A significant genetic variability was found in NSS for grain yield, days to blooming, ear and plant height, grain humidity and lodging percentage. The S-2 lines showed more frequent interaction of genotypes x environment than their S-l. In the wide sense, the heritability for the yield calculated by the analysis of variance of S-2 lines was larger than the one based on the regression of the S-2 on S-l (60 and 42%, respectively). Eight models, originated from Cockerham (1983), were used to identify the existing types of genetic variabilities. The inverse matrix method was used to estimate the parameters of genetic variability when the used co-variances gave a non-singular square matrix. The generalized inverse method o Moore-Penrose was used when the models showed a rectangular matrix. Usually, the best model was the one which estimated the additive variance only. Often times, no consistent covariance estimates were obtained among additive and dominant homocygotic (D-1) effects. For it, we could not infer to what the S-l family selection effect could be on the behavior of the resulting line crosses. The expected genetic gain per selection cycle for yield of S-2 families was 11.4%.