A grazing experiment in the seasonally dry tropics of north Queensland examined the diet selected and the growth responses of Bos indicus steers to urea supplement over two dry seasons and one wet season, from August 2001 to January 2003. There were two groups of 10 steers (control and urea-supplemented) and each group comprised two age cohorts, A and B, of five steers each with an age difference of ~1 year. In June 2002, cohort A steers were replaced with steers 2 years younger (cohort C). The steers grazed Bothriochloa pertusa pastures on a low fertility Red Chromosol soil. The groups were switched between two adjoining paddocks at fortnightly intervals when they were weighed and faecal samples were collected for faecal near infrared reflectance spectroscopy (F.NIRS) estimates of diet quality and growth rate. Rainfall and diet quality followed the expected seasonal pattern, but the 2001–02 wet season was very short (November–January) with only 65% of the long-term average rainfall. There was no rain during the 2001 dry season (DS-1) from August to mid November, and no effective rain (17 mm) between February and December 2002. Non-grass (i.e. herbaceous dicot plants) made only a small contribution to the diet, averaging 13%. In DS-1 and in the dry season of 2002 (DS-2) diet crude protein (CP) averaged 2.5% and 2.9%, and DMD/CP (ratio of dry matter digestibility to CP) averaged 18.6 and 17.1, respectively. Liveweight (LW) loss in control steers during DS-1 averaged 32 kg. Urea supplement reduced LW loss by 18 kg (P < 0.001), but most of the benefit was lost during the following growing season. During the extended dry period in 2002, average LW losses of control steers were 85 and 47 kg in the older and younger cohorts, respectively, and supplementation with urea reduced weight losses by 53 and 31 kg, respectively (P < 0.001). F.NIRS predicted the cumulative LW of the unsupplemented steers in cohorts A and B with acceptable accuracy, the difference between the observed and predicted final LW being 6 kg for cohort A after 10 months, 12 kg for cohort B after 17 months, and 27 kg for cohort C after 7 months. The results demonstrated that F.NIRS can be effectively used to monitor dietary CP and DMD levels in grazing cattle, to help develop an understanding of cattle responses to urea supplement relative to the quality of the basal forage diet, and to provide useful decision support information for the nutritional management of grazing cattle.