When are public-private partnerships (PPPs) better than conventional provision and regulated privatization? And should PPP contracts be structured and governed when this is the case?We show that the defining features of a PPP are (i) bundling of construction and operation, (ii) private but temporary ownership of assets and (iii) intertemporal risk sharingwith the public sector. Thus some characteristics of PPPs are akin to privatization while others are similar to conventional provision. Since incentives for efficient building and management are related to bundling, PPPs are closer to privatization in this regard. As the discounted government budget under a PPP is similar to that under conventional provision, PPPs are closer to conventional provision when it comes to budgetary accounting. We also show that avoiding distortionary taxation and relieving strained government budgets are weak arguments for PPPs.We examine the institutional requirements for a successful PPP program and emphasize the need for an independent supervisor of PPPs (and in general of all public works) and a Committee of Experts to award when conflicts or the need for renegotiation arises. Lack of rule of law alters the choice between conventional provision and PPPs in favor of the former, as there is less risk of regulatory takings in a short termconstruction contract than in a long lived PPP.In the case where quality service is contractible, the the PPP contract that optimally balances demand risk, user-fee distortions and the opportunity cost of public funds, features a minimum revenue guarantee and a revenue cap that differ from those observed in practice. This contract can be implemented via a competitive auction with realistic informational requirements.