The importance of migrating the South African economy from its present dependence on resource-based industries towards more knowledge-intensive economic activities has been repeatedly emphasized in many policy documents over the last decade. A higher business expenditure on research and development (BERD) is fundamental to this restructuring. Although the present level of expenditure, expressed as a percentage of total research and development (R&D) expenditure is higher than in most countries with similar levels of per capita gross domestic product (GDP), this comparison is misleading. BERD, as a percentage of industry value added, is low by international standards, especially when one considers that many of South Africa's leading industries, such as in the highly important sectors of food, metal products and petrochemicals, compete in a global market place and should be investing at levels equivalent to or even higher than their direct competitors. I analyse the structure of South African industry at sub-sector level in this paper and develop a target for manufacturing BERD, based on the industry structure and international benchmarks for such industries. As a result, I propose that manufacturing BERD should be at least R12.5 billion (an 85% increase on the 2004 values), equivalent to 0.9% of GDP (2004 rand values). I recommend that this target be incorporated into the strategic plans of the departments of Trade and Industry and of Science and Technology as an important goal for this sector and in the monitoring of their existing policy instruments, such as the recently introduced tax credit for R&D.