During the last decade, as the former system of socialist collectivized agriculture has been broken up, official statistics show that private farms and household plots have become the dominant forms of agricultural production unit in much of Central Asia and the Caucasus. In this paper we provide quantitative analysis of these new types of farming units and thus, to provide a rationale for a deeper exploration into the nature of these groups. The data used for the analysis were collected during a farm survey performed in 2001-2002 in Kyrgyzstan. The full sample consisted of 463 farms and is representative in terms of geography and different farm types. Overall we find that the total factor productivity of small groups formed on familial and social ties is higher that that of individual farms. While this suggests that familial groups are more efficient at utilising their factors of production, non-parametric estimation illustrating the relationship between production, land and labour, suggests that these groups are not optimising production as they appear to be operating under increasing returns to scale. A parametric production function estimation and an analysis of total factor productivity provide further support of this. In conclusion, some discussion is provided to explain why groups may be operating under different returns to scale technology than individual farms. These explanations hinge primarily on understanding the asset-pooling, risk-sharing and labour specialisation functions of groups.