In contrast to the purines, the routes that have been proposed for the prebiotic synthesis of pyrimidines from simple precursors give only low yields. Cytosine can be synthesized from cyanoacetylene and cyanate; the former precursor is produced from a spark discharge in a CH4/N2 mixture and is an abundant interstellar molecule. But this reaction requires relatively high concentrations of cyanate (> 0.1 M), which are unlikely to occur in aqueous media as cyanate is hydrolysed rapidly to CO2 and NH3. An alternative route that has been explored is the reaction of cyanoacetaldehyde (formed by hydrolysis of cyanoacetylene) with urea. But at low concentrations of urea, this reaction produces no detectable quantities of cytosine. Here we show that in concentrated urea solution--such as might have been found in an evaporating lagoon or in pools on drying beaches on the early Earth--cyanoacetaldehyde reacts to form cytosine in yields of 30-50%, from which uracil can be formed by hydrolysis. These reactions provide a plausible route to the pyrimidine bases required in the RNA world.