Analysing the first decade of post-communist pension policy in Hungary, Poland and the Czech Republic, we find an interesting policy pattern: while the Czech government has implemented a pension reform characterised by gradual and partially parametric policy changes, Hungarian and Polish policymakers have enacted a far-reaching reform of their countries' old age security systems, involving a transformation of the previous one-pillar system into a multipillar system based on a substantial fully-funded element. Taking up the neo-institutionalist literature, which conceives pension policy as 'locus classicus' for path-dependent policy development, the central analytical question arising is why Hungary and Poland on the one hand and the Czech Republic on the other took so different reform paths during the 90s - despite the fact that all three countries were characterised by rather similar pension systems under communist rule. By referring to the theoretical framework provided by the multiple streams approach of John W. Kingdon, the paper develops an explanatory argument which focuses on the role of the World Bank in the agenda-setting/pre-decision phase of the pension reform processes throughout the three Visegrád countries. In a nutshell, the paper argues that the rapidly increasing problem pressure within the Hungarian and Polish pension system during the mid-90s finally opened a policy window for a far-reaching pension reform, which allowed the policy entrepreneur World Bank to couple its elaborated country-specific pension proposals to both the respective problem and policy streams. In the Czech Republic, in contrast, the crucial policy window did not open.