The present paper argues that the effect of corruption on foreign ownership is not necessarily linear and depends on the level of host corruption. So long as the expected returns from foreign investments exceed its expected costs, higher host corruption will be associated with higher foreign ownership. However, costs may exceed or exactly compensate the returns to foreign investment at very high level of corruption, giving rise to negative or even an insignificant relationship when positive and negative effects outweigh each other. Further, we argue that this non-linear corruption effects may arise from multinational firms' attempts to investing in countries with similar environment and/or ensuring some formal networking with host countries in a bid to limit the damages caused by high level of host corruption. Panel fixed effects estimates (after correcting for foreign entry selection) using a recent large home-host matched panel data from central and eastern European host countries provide some support to these hypotheses: (i) higher corruption is associated with significantly higher foreign ownership unless corruption is at its fourth quartile value. (ii) There is also some confirmation that this non-linear corruption effects is linked to parent firms' attempts to ensure institutional similarity while investing in corrupt host countries: in particular, foreign multinationals from EU/OECD home countries tend to hold higher ownership in EU/OECD host countries and also when the home-host relative corruption distance is small.