The inverse relationship between farm size and productivity has almost become a 'stylised fact' in the economic development literature. Most of the studies contributing to this preception have been flawed by methodological shortcomings and the request is that these studies be treated with caution. Using recent farm survey data from the wine producing areas of the Western Cape of South Africa, this study attempts to overcome some of the methodological problems, distinguishing between partial and total productivity measures. Using data envelopment analysis, most of the wine grape producers were found to operate under constant returns to scale. Co-operative membership seemed to overcome the economies of scale associated with processing and marketing. The inverse relationship between farm size and both land productivity and total factor productivity is weak, not consistently negative and differs between regions. Thus, caution must be used when advocating rural development policies based on the inevitability of an inverse relationship existing in all sectors and production regions of agriculture. © 1998 Elsevier Science B.V. All rights reserved.