Experiments were carried out on the exhaust of an engine on a test bed with a selective catalytic reduction catalyst. The reductant could be introduced either as 5% ammonia in nitrogen gas or as a spray of aqueous urea droplets. Conversion of nitrogen oxides was investigated at temperatures near 488 K (215 °C) and 533 K (260 °C), which are typical of a diesel passenger car exhaust system. The experiments with a urea spray were carried out for three reductant levels, i.e. for two reductant levels that were deficient and for one reductant level that was approximately stoichiometric. The experiments were run to steady state conditions. The spray was introduced into the system in two different ways, sprayed either into a mixing can and nozzle arrangement or into the system via an oblique pipe; in the latter case the spray impinged on an angled plate mixer. A transient case was also investigated where the engine load was ramped up from a brake mean effective pressure of 6 bar to 10 bar over 20 s and, after reaching the steady state, was ramped back down to a brake mean effective pressure of 6 bar over 20 s and allowed to reach a steady state. The nitrogen oxide supplied was 100% nitric oxide in all these experiments. This was achieved by using a palladium diesel oxidation catalyst to remove hydrocarbons and carbon monoxide from the exhaust stream but without oxidising the nitric oxide. Measurements were made with a Fourier transform infrared gas analyser. The steady-state results showed that the percentage nitric oxide conversion observed using a urea spray had a value that was about 10% below the percentage conversion observed with ammonia gas when using either spray configuration. There was evidence that urea droplets were being transported unconverted through the selective catalytic reduction catalyst in both steady-state engine-load tests and transient engine-load tests. The ammonia deficit was 20% or more of the potential amount of ammonia injected by the aqueous urea spray.