After three decades of expulsion, the industrial policy (IP) is again a part of legitimate policy agenda in developed and emerging economies alike. The theoretical case for IP interventions remains valid in the presence of continued market failures enhanced by growing share of international trade and stronger connectedness in the global economy. But dangers of government failures are equally present, either through errors of commission (due to excessive and wrong government intervention)or omission (failure to act and correct the obvious market failures). At the practical policy level the issue is no longer 'why' IP is needed, or whether the government should engage in IP, but 'how' to design and implement IP measures that would avoid the known pitfalls of the past and help sustain development and obtain desirable economic restructuring. Recent debates (Lin-Chang, Rodrik-Lerner) and thorough literature surveys(Harrison)confirm the need for IP, but recommend a close alignment with comparative advantage (Lin), shift from hard (tariffs, subsidies) to soft IP interventions aimed at increasing productivity when ever possible (Harrison), and suggest avoiding risks of poorly designed IP measures through joint government-private sector 'self-discovery' of optimal IP policy parameters not known ex ante (Rodrik). Advanced WTO and bilateral trade arrangements may seriously limit the scope for legitimate new industrial policy.