The effects of the field production environment on the growth, development, seed yield and quality of soybean (Glycine max [L.] Merrill) were investigated on a Wakanui silt loam soil in Canterbury (Lat. 43° 38' S and Long. 172° 30' E), New Zealand. A sowing date trial was conducted in two seasons (1999/2000 and 2000/2001) and a plant density trial in 2000/2001 only. In 1999/2000, four cool tolerant soybean cultivars (Northern Conquest, Maypole, Alta and March) were sown at three dates (15 November, 7 December and 29 December 1999), with each cultivar/sowing date treatment being replicated three times. The experiment in 2000/2001 used two cultivars (Northern Conquest and March) with four sowing dates (2 October, 17 October, 1 November and 16 November 2000) and four replicates of each treatment. A split-plot design was used in both experiments with sowing date being the main plot and cultivar being the sub-plots. The plant density trial used two cultivars (Northern Conquest and March) sown on 2 November 2000 in a systematic (radial) design which offered an opportunity to test a range of plant densities from 3.1 to 71.6 plants m⁻². The cultivars and plant densities were replicated four times. Soybean crop development from sowing to emergence was strongly related (r² = 0.91) to soil temperature, and from flowering (R1 stage) to seed physiological maturity (R7 stage) to air temperature (r² = 0.88 - 0.95), but from emergence to flowering was related to both air temperature and photoperiod (r² = 0.95). The pod set stage (flowering to start of seed fill, R5 stage) was also related to temperature and photoperiod. Seed yield was strongly related with crop growth rate (CGR) (r² = 0.86) and total dry matter (TDM) at R5 stage (r² = 0.73). Crop growth rate (CGR) was significantly related to TDM at the R5 stage (r² = 0.71). Seed growth rate was also highly associated with CGR (r² = 0.91). The TDM produced at the R5 stage is the function of CGR and time. CGR depends on the amount of photosynthetically active radiation (PAR) intercepted by the canopy. Therefore, high TDM production at the R5 stage or at harvest depends on the attainment of canopy closure (interception of >95% incoming solar radiation) by the early reproductive stage. A green area index (GAI) of 3.4-3.5 was required to attain canopy closure. Early canopy closure and a higher PAR interception were possible by sowing the crop between mid October and mid-November. This helped produce higher TDM at the R5 stage, which determines the most important yield components - seed number per square meter and individual seed weight. Seed yield ranged between 249 and 300 g m⁻² in these cold tolerant cultivars when the crop was sown in October or November. However, December sown crops gave a very poor seed yield (53-178 g m⁻²), mainly because they did not reach canopy closure and consequently produced lower TDM due to increasing temperature and decreasing photoperiod. This environment enhanced phenological development, reduced branch and leaf development and ultimately allowed less total PAR interception. Cv. March out yielded cv. Northern Conquest by about 13-20%. However, the seed yield of cv. Maypole and Alta did not differ from that of cv. March. There was an asymptotic relationship between seed yield and plant density and a plateau in seed yield was obtained above a plant density of 52.3 plants m⁻². The seed yield for cv. Northern Conquest and March at 52.3 plants m⁻² was 330 g m⁻² and 437 g m⁻². There was a strong relationship between seed yield and TDM produced at around the R5 stage (r² = 0.76), intercepted PAR (r² = 0.90) and CGR (r² = 0.64). There was a significant correlation between plant density and TDM (r = 0.75***) and seed yield ((r = 0.67**). This was because increased plant density helped early canopy closure and increased PAR interception. A plant density of 52.3 plants m⁻² or more was required to obtain high soybean yield of these cultivars in Canterbury. Seed yield was strongly related with TDM at harvest (r = 0.93*** - 0.96***) in both sowing date and plant density trials. However, the relationship between seed yield and HI was poor and inconsistent. The cultivars used in the trials had a seed protein content ranging between 34 and 42%. Seed oil content ranged between 12 and 15%. Sowing date had no influence on seed protein content, although seed protein content showed a curvilinear relationship with plant density. Seed coat cracking is primarily a genetically controlled character that was increased by a cold environment. Both cv. Alta and Maypole were particularly prone to seed coat cracking induced by cold temperatures during seed development. Sowing between mid-October and mid-November produced seeds with high germination and vigour. This was because the environment during the vegetative stage was favourable for high crop growth and subsequently adequate assimilate supply during seed development. Although high seed yield and quality in soybean were possible for these hand harvested seeds with sowings between mid-October and mid-November, the mid-November sowing is not considered suitable for high quality seed production because the crop matured in April when the environment became cool and machine harvesting of the crop at an optimum seed moisture content of 15% was not possible. Therefore, mid-October to early November was the optimum sowing date for these soybean cultivars in Canterbury. Plant density and pod position had no effect on germination percentage of soybean seed. Seed germination was high (>92%) at all plant densities and also at the three pod positions (top, bottom and branch). However, seed vigour was inversely related to plant density, decreasing as plant density increased. Irrespective of plant density, seeds from the bottom position (up to node 8) on the plant had the highest vigour, followed by seeds from the branch pods, while seeds from the top pods had very poor vigour.