Abstract: This study on path coefficient and polynomial model assessed the direct and indirect contributions of upland rice characters. A mixed of stepwise regression procedure in SAS Insti-tute with a probability to enter and remove variables was set at 0.05, and was used to select the 13 yield-related traits. The results revealed that there were direct effects on paddy yield in which number of grains contributed the highest, followed by percent filled grains and 1000 grain weight while indirect contribution of growth and yield components via number of tillers, leave area index and 1000 grain weight were negative.The individual percent contribution showed that 1000 grain weight contributed to the yield more than any other parameters followed by number grains per panicle while the least individual percent contribution was made by panicle weight. The highest indirect percent contribution to yield via one another were observed through number of filled grains per panicle via number of grains per panicle, followed by number of grains per panicle via 1000 grain weight and the least was plant height via number of tillers. The unaccounted residual effect constitutes 9.13% and 5.55%. The polynomial response showed that optimum seed rate of 85.7 kg ha-1 was observed with quadratic function of Y=-0.2765X2 + 47.409X-36.4 while in 2012, the optimum seed rate of 81.3 kg ha-1 was recorded with quadratic function of Y=- 0.3044X2 +49.491X + 2252.8. The optimum seed rate of 83.4 kg ha-1 with quadratic function of Y=- 0.2905X2 +48.451X +1108.1 was recorded for the combined.