This paper investigates pair wise spatial integration and efficiency for central maize markets in three southern African countries: Malawi, Mozambique, and South Africa. The goal is to evaluate the extent to which cross border trade restrictions on maize trade affect market integration and efficiency, vis-à-vis related costs of transfer. We also attempt to establish the most dominant form of efficiency barrier in those markets. The analysis employs the extended parity bounds model (Barrett and Li 2002), supported by non-parametric assessments; and uses monthly wholesale on prices, trade flow, and estimated transfer costs for the period 1996 to 2004. Results reveal significant frequency of market integration between markets in close proximity, regardless of country location, indicating tradability of commodities and contestability of markets across borders. For those markets, however, efficiency appears weak, as trade often fails to exhaust arbitrage profits. We observe inefficiency driven by insufficient arbitrage at least 60% of the time. Occasionally, trade is observed in the presence of negative returns, though at a lower frequency of about 10%. Markets not linked through trade tend to have a higher frequency of efficiency (up to 52%), so that the lack of trade often is justified by the lack of positive arbitrage returns. In those cases, market segmentation appears driven more by restrictive transport costs than cross-border trade barriers. Our results suggest that the dominant forms of inefficiency in the southern African maize markets considered in this study are (1) insufficient arbitrage resulting from supply side constraints and other non-cost barriers to trade, and (2) restrictive transport costs. Border administered tariffs and other forms of taxes on imports seem to account for a relatively low proportion of transfer costs, and generally reduce arbitrage returns marginally.