Lentils (Lens culinaris Medik.) were sown on eight sowing dates from April to November in two seasons in Canterbury, New Zealand. In 1984/85, six sowing dates were combined with two lentil cultivars (Olympic and Titore) and two irrigation treatments. In 1985/86, Titore was sown on two dates, with four irrigation treatments. An additional experiment grown under rain shelters examined the response of Titore to four irrigation regimes. The 1984/85 season was dry and rainfall was only 70% of the long-term mean. In this season, seed yield was high, 33 t/ha from the May sowing. The 1985/86 season was wetter than average and seed yields were lower, ranging from 0-6 to 1-5 t/ha. Under rain shelters, seed yield ranged from the equivalent of 0-32 to 2-5 t/ha.Sowing date had the most marked effect on seed yield. In the 1984/85 season, all autumn and winter sowings yielded 2-4-3-3 t/ha, whereas the spring sowings yielded 0-5-1-5 t/ha. In 1985/86, unirrigated plots from the May sowing yielded 1-5 t/ha, whereas all other plots yielded c. 0-8 t/ha.Generally, the small-seeded cultivar Titore outyielded Olympic. Dry matter (DM) accumulation followed similar trends to seed yield. Seasonal DM accumulation followed a sigmoidal curve. Functional growth analysis indicated that plants from autumn/winter sowings had a weighted mean absolute growth rate of 110-171 kg/ha per day, whereas spring-sown plants grew at 96-137 kg/ha per day. The maximum crop growth rate was 230 kg/ha per day in the July 1984 sowing.There was little positive response to irrigation in both seasons. Under rain shelters, there was a linear increase in both dry matter and seed production with increased total water. Fully irrigated plants produced 1-27 g DM and 0-72 g seed/m s per mm of water received.In the field experiments there was no relationship between maximum potential soil moisture deficit (D) and yield. Under rain shelters, however, there was a linear relationship which indicated a limiting deficit of c. 130 mm. The relationship showed that, for each millimetre increase in D above O,, 0-39% of the maximum yield was lost.Under the rain shelters, there was a strong relationship between yield and actual evapotranspiration (ET). Water-use efficiency (WUE) ranged from 2-81 g DM/m 2 per mm ET in unirrigated plots to 0-69 g seed/m 2 per mm ET.