Due to changes in the global agricultural system and support from various organizations, contract farming has recently been significantly expanded in many developing countries. A considerable body of literature analyses the impact of contract farming on the welfare of smallholders, whereas its impact on efficiency and productivity is mostly overlooked. This study addresses this salient gap by combining the approaches suggested by Bravo-Ureta, Greene, and Solís (Empirical Economics 43:55–72, 2012) and Rao, Brümmer, and Qaim (American Journal of Agricultural Economics 94:891–912, 2012). We first use the approach of Bravo-Ureta, Greene and Solís (2012) to estimate two separate production frontiers (one for contract farmers and one for non-contract farmers) that account for potential biases due to self-selection on both observed and unobserved variables. Then, we follow Rao, Brümmer and Qaim (2012) and create a meta-frontier in order to estimate the effects of participation on the farms’ meta-technology ratio, their group technical efficiency, and their meta-technology technical efficiency. The empirical analysis uses a cross-sectional data set from sunflower farmers in Tanzania, where some of the farmers participate in contract farming while others do not. We find a significant selection bias, which justifies the use of the sample selection framework. Our preliminary results indicate that contract farming significantly increases the yield potential (meta-technology ratio) but lowers the group technical efficiency. As the first effect is slightly larger than the second, we find a small positive effect of contract farming on productivity (meta-technology technical efficiency). The positive effects on the yield potential and the (average) productivity can be (at least partly) explained by the contractor’s provision of (additional) extension service and seeds of high-yielding varieties to the contract farmers.