Fire often increases the productivity of perennial tussock grasslands in mesic environments but can reduce growth for one or more growing seasons in arid and semi-arid environments. We examined effects of single-burns on growth and nutrient content of grasslands in sub-tropical, northwestern Australia. These grasslands were dominated by Themeda triandra, a species often managed by regular burning in wetter temperate and tropical zones. Burns were in the late dry season and were replicated using small plots (5 x 5-m) within fenced areas at two sites. Total projective cover and aboveground biomass were significantly less in burnt plots relative to controls for 2.5 years after burning despite four growing seasons, including the first summer, of above-average rainfall. Recovery of burnt plots was hindered by an extended dry period in the second year, demonstrating that rainfall in subsequent seasons can be as important as rainfall in the first season in determining post-burn productivity of grasslands in semi-arid environments. Greater decreases in grass cover in burnt plots during the extended dry period may have been due to less standing dead and litter than controls, and therefore less insulation from extreme summer temperatures, although relationships between cover changes and cover at the start of the period were weak. With the exception of increased pH near grass tussocks, burning had little effect on chemical characteristics of surface soils in the first week. Concentrations of N, and particularly P, in aboveground plant material were greater in burnt plots four months after burning, following summer rains, but were either less than or similar to those in controls with increasingly dry conditions. Significantly lower concentrations of P in green foliage from burnt plots during dry seasons, when uptake from soil pools would be minimal, indicated that burning decreased P retranslocation from plant stores. However, we found no evidence that single-burns increased nutrient limitations to growth because plant contents of N and P were comparable in burnt and control plots during periods of adequate water supply. Our data support previous generalizations that prescribed burning of perennial tussock grasslands in semi-arid environments is mostly unnecessary because putative benefits of increased productivity and forage quality, characteristic of more mesic environments, were not realized.