Studies at Bushland, TX in 2002, 2003, and 2004, and at Tribune, KS in 2004 compared different planting geometries on production of dryland grain sorghum (Sorghum bicolor L. Moench). The hypothesis was that grain sorghum in clumps compared to uniformly spaced plants would produce fewer tillers and less early vegetative growth. Therefore, more soil water would be available during the critical reproduction and grain filling periods. Growing conditions during 2002 and 2003 were extremely challenging because of below-average and erratic precipitation. However, grain yields were as much as 100% higher for grain sorghum grown in clumps than for conventionally seeded plants. In 2004, growing season precipitation at Bushland was quite favorable and it greatly exceeded normal amounts at Tribune. All treatments were hand seeded in 75-cm rows. Within-row treatments were: SP-25, single plants every 25 cm; SP-38, single plants every 38 cm; SP-25-TR, single plants every 25 cm with tillers removed by hand; C3-75, clumps of 3 plants every 75 cm; and C4-100, clumps of 4 plants every 100 cm. All treatments had a density of 54,000 plants ha-1, except SP-38 that had 36,000 plants/ha. Tiller formation was significantly less for the C4-100 and the C3-75 treatments and this resulted in much less early season biomass production. Clump treatments headed earlier and yields were about 35% higher when yields were in the 2500 kg/ha range and about 25% in the 3500 kg/ha range. There was no increase where yields approached 5000 kg/ha. At Tribune, there was a slight, but statistically significant, decrease for the C4-100 treatment where the yields were more than 6000 kg/ha. There were apparently not enough heads to support this yield level. This was also borne out by the SP-25-TR treatment that produced significantly lower yields when yields exceeded 3000 kg/ha.