Methicillin-resistant Staphylococcus aureus (MRSA) strains are important nosocomial pathogens worldwide and now are also of growing importance in community-acquired infection. Their resistance depends upon a supplementary peptidoglycan transpeptidase, PBP2' (PBP-2a), which continues to function when normal PBPs have been inactivated by ß-lactams. PBP2' is encoded by the mecA gene, which is carried by the staphylococcal cassette chromosome, a large and somewhat variable DNA insert of uncertain origin. PBP2' does not wholly lack affinity for ß-lactams, but its affinity for available analogues is very weak. In principle, it should be possible to re-engineer ß-lactams to bind PBP2' strongly, and the desirability of this approach is self-evident: no other antibiotic class has a record equal to the ß-lactams for safety and efficacy. Moreover, there is consensus that ß-lactams are inherently more efficacious than vancomycin against infections due to susceptible staphylococci. In practice, finding viable PBP2'-active ß-lactams has proved difficult and the catalogue of near-misses extends back to the 1980s. At last, however, one cephalosporin with high affinity for PBP2'—ceftobiprole—is entering phase III trials. Ceftobiprole inhibits MRSA at 1–2 mg/L under standard conditions. Even when mecA/PBP2' was induced strongly, ceftobiprole MICs for MRSA only reached 4 mg/L, a clinically attainable concentration. A phase II trial in skin and skin structure infection recorded cures by ceftobiprole in 4/4 MRSA infections, and results of the phase III trials are awaited with great interest.