The divergent growth rates of New Zealand regions are a matter of concern to economists, geographers and policy makers. People have been concerned about the 'drift to the north' since at least the 1920s. Geographers have paid particular attention to the role of manufacturing industry in explaining the growing hegemony of Auckland (Taylor and Hosking, 1979), and in generating regional inequalities through differential wage payments (Hoare, 1986). They have also analysed the regional development policies put in place at various times in response to concerns about these divergent growth rates (Le Heron, 1979; Scott, 1980). This literature largely overlooks a significant cause of regional growth and decline. The omitted cause is industry protection policies such as import licensing, tariffs, and export subsidies. Scott (1980) hinted at the role that these policies play in creating regional imbalances but provided no supporting analysis. Some regional implications of the 1980s process of protection reform were noted by Patterson (1989) but it was St Hill (1987) and Treasury (1987) who provided the most comprehensive analysis. These studies found that Auckland, Lower Hutt, Wellington, and Christchurch were favoured by having a high concentration of protected, import-substitute manufacturing. Provincial regions were less favoured because their industries tended to be export oriented ones which received lower levels of assistance. The aim of this paper is to describe further the regional costs and benefits of industry protection policies. These costs will be measured in monetary and employment terms under a variety of assumptions. The results suggest a paradox. Regions suffering the heaviest burden under industry protection policies were generally 'priority regions' for enhanced development. The penultimate section of the paper gives some reasons for this paradox. First, however, I outline ways in which assistance to industries and regions can be quantified. Then regional rates of assistance during the 1980s are measured. The earliest available data (1981/82) are used to make inferences about the pattern of costs and benefits before the current liberalisation. The initial impact of the liberalisation is assessed using data from the late 1980s.