In an experiment carried out at Itararé Experimental Station, a standard height cultivar IAC-5 was crossed with the semidwarf cultivar Siete Cerros and the dwarf cultivars Tordo, Vican-71 and Olesen. Parents, F1s , F2's and reciprocal backcrosses were tested for grain yield, plant height, number of spikes per plant, number of spikelets per spike, number of grain per spike, number of grain per spikelet, 100 grain weight, spike lenght, rachis internode lenght. All data were determined on an individual plant basis. Broad sense heritability estimates were very high for plant height, moderate for number of spikelets per spike, rachis internode lenght, number of grain per spikelet, number of spikes per plant, number of grain per spike and 100 grain weight, low for grain yield and spike lenght. Narrow sense heritability estimates were very high for plant height and moderate to low for the rest of agronomic characteristics under study. Additive effects were the main source of genetic variation for all studied characters except for 100 grain weight. Plant height was significantly correlated in all studied populations with grain yield, number of spikes per plant, number of spikelets per spike, number of grain per spike, 100 grain weight and spike lenght. Segregating population from the crosses IAC-5/Tordo and IAC-5 showed no correlations between plant height and number of grain per spikelet and no associations between plant height and rachis internode lenght was verified for the F2 populations from the crosses IAC-5 /Vican-71 and IAC-5/Olesen. The results showed that any source of dwarfism could be used efficiently in a breeding program towards the development of semi-dwarf associated with high yield potential lines. However larger F2 populations would be required to ensure the frequency of desired recombinants.