Seven field trials were conducted at Lincoln to examine the relationships between lucerne density, lucerne yield, and weed invasion. In five trials seeding rates from 0.5-36 kg/ha were compared and two trials were established by planting with densities of 8-140 plants/rn-. Establishment and death rates were density-dependent, and densities converged with time to a level above that required for maximum production. The density-yield relationship reached a plateau at 30 plants/m-, although higher densities were more productive in the first season. The difference between sparse and dense stands was less in summer than in other seasons. Annual weed production, which was inversely proportional to stand density, partially compensated for low production in sparse stands. It was concluded that with proper care farmers could use a seeding rate of 2 kg/ha, which is much less than rates currently recommended. MATERIAL AND METHODS "self-thin" to an equilibrium density above the minimum required for maximum production (Fribourg & Kennedy 1953; Zaleski 1959; Yoda et at. 1963; Takasaki et at. 1970). Thus. there is only a temporary advantage in sowing more seed than will give this density initially. In New Zealand lucerne seed is usually sown at 8-10 kg/ha or about 300-500 seeds/m''. This may be more than sufficient to give maximum yield for a maximum period, or for the planned life of the crop. Trials were conducted over several years to examine the effects of sowing lucerne at varying seeding rates and with various cover crops. In this paper the relationships between plant density, yield, and weed invasion are discussed.