Nitrapyrin and carbon disulphide (CSz), or sodium or ammonium trithiocarbonate (which evolve CSZ in soil), were injected with aqueous ammonia (NH3) into grassland in three successive seasons at Rothamsted and at Woburn, using a new site each season. Injection was done either in November ('autumn'), or in the following February or March ('spring'). When applied in autumn, the inhibitors increased yield and %N of grass after the mild, wet winter of 1974-75, but had little or no effect in the two subsequent seasons. These effects are consistent with rates of nitrification of the injected NH3, estimated by KCl extraction of soil taken from injection slits, and with winter rainfall. When inhibitors were applied in spring, yields were consistently improved at Rothamsted but depressed at Woburn. All effects were small. Nitrapyrin or CS2, injected with aqueous NH3 into grassland in autumn, reduces nitrogen losses that occur in winter. However, judging from grass yields, in these experiments (in south-east England) losses were only a small fraction of the dose of N applied.Injecting these inhibitors with NH3 in spring can delay nitrification and increase the persistence of the injected N, but this will not necessarily increase yield.