Competition between rice (Oryza sativa) and Echinochloa crus-galli (L.) Beauv. (barnyard-grass or cockspur) was studied experimentally in the field in Sri Lanka. By growing rice in bags sunk in the paddy soil such.that the roots of the plants were either separated from or free to mingle with those of neighbouring weeds, or by growing rice in the same bags but in the absence of weeds, it was possible to calculate the relative importance of interference below ground (root competition) and above ground (shoot competition). Three rice varieties with different above-ground morphology were inhibited to different extents by E. crus-galli but in each case root competition was more important than shoot competition. In the variety that was most sensitive to root competition, Bg 400–1, inhibition of total growth due to root competition increased with increasing weed density. When grown in monocultures in pots in the field, the two species exhibited similar growth rates whether or not the soil was fertilized, but in 1:1 mixtures growth of E. crus-galli was greater, particularly if fertilizer was added, as this promoted the growth of the weed, while having no effect on the rice. In mixture, relative yield totals were close to 1.0, with the relative yield of E. crus-galli being consistently greater than 0.5, particularly with added nutrients, while the relative yield of rice was always less than 0.5. The relative crowding coefficient for rice with respect to E. crus-galli (Keb), when based on total plant dry weight, was lower in fertilized than in non-fertilized soil, falling in the former case from 0.4 to 0.2 during the experiments Krootrb, based on root dry weight, decreased with increasing soil depth and was lower in fertilized than in non-fertilized soil, most notably after 62 days of growth, when root dry weight was at its maximum. It is concluded that inhibition of root growth of rice, leading to a reduced ability to obtain resources from the soil, was the major factor contributing to the decline in the growth of rice in the presence of E. crus-galli.