Flash vacuum pyrolysis of 2-allyloxypropenoic esters (e.g. 7) gives benzo[b]furans (e.g. 32) in synthetically useful yields by sequential generation of a phenoxyl radical, cyclisation and ejection of the carboxylic ester function as a free radical leaving group. The method is compatible with a range of substituents on either the benzene ring or the propenoate chain, and is particularly effective for 2-substituted benzo[b]furans. The natural products 5-methoxybenzo[b]furan 1 and angelicin 2 have been synthesised in three and four steps respectively from commercially available starting materials by this route. Related cyclisations to give naphtho[2,1-b]furan 40 were complicated by competitive formation of naphtho-[2,1-b]pyran-3-ones (e.g. 41 and 42), but the yield of the required product could be optimised by the choice of the radical precursor. Annelation of a furan ring onto a thiophene is also possible by this method, but lower yields are obtained in such pyrolyses.In earlier papers we have reported application of the technique of flash vacuum pyrolysis (FVP) to the gas-phase generation of aryloxyl radicals, and their reactions with adjacent aromatic systems. 2,3 These processes are often dominated by hydrogen abstractions, and relatively small amounts of useful cyclisation products are formed. In extending this work to the properties of aryloxyls with adjacent alkene systems, it became apparent at an early stage that the carboxylic ester function behaves as a specific and highly efficient radical leaving group under these conditions, leading to a useful synthetic route to benzo[b]-furans. 1 We now present full details of this work, and its extensions, which have led to concise syntheses of the natural products 5-methoxybenzo[b]furan 1 and angelicin 2. The results of preliminary attempts to annelate a furan ring onto fivemembered heterocyclic systems are also reported.As in our previous work, 2,3 we have employed O-allyl or Obenzyl ethers as radical generators, made in high yield from the corresponding phenol by treatment with the appropriate bromoalkane in dimethylformamide containing anhydrous potassium carbonate. The alkene function was made from an appropriate carbonyl compound by Wittig or Knoevenagel methodology (Scheme 1). In many cases, the order of the allylation and Wittig steps was unimportant, though in practice 2-allyloxybenzaldehyde 3 was used to make the Wittig products 4 and 14 (obtained in low, but unoptimised yields) and the Knoevenagel products 5, 19 and 20 (61-97%). Compounds 5 and 20 were made from the appropriate active methylene compound using piperidinium acetate catalyst, but it proved necessary to use more vigorous conditions (titanium tetrachloridepyridine) for the synthesis of the diester 19. The precursors 7, 9, 11, 13, 16 and 18 were made by allylation of the phenols 6, 8, 10, 12, 15 and 17 (80-100%) respectively, which were themselves made by Wittig reactions with salicylaldehydes, to give predominantly-often exclusively-the E-isomers (75-90%). 4 The alkoxynaphthols 22-24 and 26 we...