A field experiment was conducted on an acid sulfate soil in Thailand to determine the effect of N fertilization practices on the fate of fertilizer-N and yield of lowland rice (Oryza sativa L.). A delayed broadcast application of ammonium phosphate sulfate (16-20-0) or urea was compared with basal incorporation of urea, deep placement of urea as urea supergranules (USG), and amendment of urea with a urease inhibitor. Deep placement of urea as USG significantly reduced floodwater urea- and ammoniacal-N concentrations following N application but did not reduce N loss, as determined from an15N balance, in this experiment where runoff loss was prevented. The urease inhibitor, phenyl phosphorodiamidate (PPD), had little effect on floodwater urea- and ammoniacal-N, and it did not reduce N loss. The floodwater pH never exceeded 4.5 in the 7 days following the first N applications, and application of 16-20-0 reduced floodwater pH by 0.1 to 0.3 units below the no-N control. The low floodwater pH indicated that ammonia volatilization was unimportant for all the N fertilization practices. Floodwater ammoniacal-N concentrations following application of urea or 16-20-0 were greater on this Sulfic Tropaquept than on an Andaqueptic Haplaquoll with near neutral pH and alkaline floodwater. The prolonged, high floodwater N concentrations on this Sulfic Tropaquept suggested that runoff loss of applied N might be a potentially serious problem when heavy rainfall or poor water control follow N fertilization. The unaccounted-for15N in the15N balances, which presumably represented gaseous N losses, ranged from 20 to 26% of the applied N and was unaffected by urea fertilization practice. Grain yield and N uptake were significantly increased with applied N, but grain yield was not significantly affected by urea fertilization practice. Yield was significantly lower (P = 0.05) for 16-20-0 than for urea; however, this difference in yield might be due to later application of P and hence delayed availability of P in the 16-20-0 treatment.