Using the variability present in the F2 population originating from six crosses of rice mutant strains, selection was
practiced for high grain yield. Expected response to selection was estimated in F3 generation and compared with F2
generation. Similar selection practices were made in F3 generation and selection response was estimated in F4 generation.
F3 generation showed significant positive selection response for plant height, number of panicles per plant, grain density
and days to 50 % flowering, thus indicating the effectiveness of selection for these characters. Grain yield showed non
significant response in F2 generation which indicates straight selection for this character during early generation may not
be effective. F4 generation showed positive selection response for all the characters i.e. plant height, number of panicles
per plant, grain density, days to 50 % flowering and grain yield. This indicates the effectiveness of selection for these
characters in F3 or latter generation. The regression coefficient between F2 and F3 generation and F3 and F4 generation was
significant for all the characters. Selection for plant height, number of panicles per plant and days to 50 percent flowering
can be made in the early segregating generation. Grain yield per plant and grain density is effective for selection in later
generation.