Preparations of the enzyme isopenicillin N synthetase from Cephalosporium acremonium CO 728 convert the modified tripeptide substrate 6-(L-α-aminoadipoyl)-L-cysteinyl-D-allylglycine into penam, homoceph-3-em, hydroxycepham, and hydroxyhomocepham type products; these have been isolated and their structures established. We now report that the conversion of the similarly modified tripeptide 6-(L-α-aminoadipoyl)-L-cysteinyl-D-allylglycine (lc) with this purified enzyme at pH 7-8 leads simultaneously to six distinct β-lactam containing products (2c), (2d), (3b), (4a), (4b), and (5). The first product, vinylpenicillin (2c), showed penicillin-type antibacterial activity against Gram-positive organisms (Staphylococcus aureus N.C.T.C.6571, Sarcina lutea 400E, Bacillus megaterium 1571E) but no effect on Gram-negative species (Alcaligenes faecalis DS.367, Pseudomonas aeruginosa N.C.T.C.10701, Escherichia coli 2110E, Salmonella typhi DS.48) at a concentration of 50 μg ml⁻¹. The second class of products, homoceph-3-ems (4a,b) [ratio (4a):(4b) = 10:1], had (4a) showing no antibacterial activity against S. aureus N.C.T.C.6571 at 400 μg ml⁻¹ and equilibrating with (4b) in buffer. The third class was hydroxymethylcepham (3b) and the fourth was hydroxyhomocepham (5), both showing no antibacterial activity. In summary, the diverse structural range of products formed from the allylglycine tripeptide (lc) with isopenicillin N synthetase may be explained by two discrete pathways (routes a, b) from a common monocyclic β-lactam intermediate (6). From route a, a radical (or carbon-metal) type intermediate (7) would lead, on ring closure, to the penicillin products (2c) and (2d) whilst the rearranged radical (8) would give the homoceph-3-em (4a). The preference of (2c) over (2d) has precedent in a related system, whilst the predominance of (4a) over (2c) and (2d) probably reflects a lower energy barrier for the second ring closure. The product (4b) probably arises by epimerisation of the biosynthetic (4a). The novel hydroxylated products (3b) and (5) are thought to arise by route b. Whether the source of the oxygen atom [e.g., at C(3) of (5)] is the co-substrate dioxygen or, alternatively, water is currently under investigation.