Smallholder livelihoods in agricultural areas in Tibet Autonomous Region, China (Tibet for short) have traditionally been based on subsistence mixed crop-livestock systems. Like many parts of China and the developing world, rural Tibet is undergoing rapid change in agricultural development, boosted by links with the off-farm sector. However, the agricultural transition process in Tibet has not been linear, is affected by policies particular to the region, and varies significantly for different categories of rural actors. This paper analyses heterogeneity of household types in agricultural areas of southern Tibet and how household structures and characteristics result in different agricultural development pathways. Data is drawn from a stratified random survey of 144 households in three townships in southern Tibet and analysed through three methods: cluster analysis, bio-economic modelling, and analysis of household perceptions and attitudes to change. The analysis allows for the identification of three key household types — semi-subsistence, pluriactive and semi-commercial — organised along a continuum of intensification, commercialisation, specialisation and productivity. The study demonstrates the diversity of household farming systems in Tibet which in turn highlights the need for disaggregated analysis and tailored development policies and strategies. However, the analysis also reveals commonalities in development paths between groups, where all farm types are more interested in pursuing income and livelihood goals through an increase in productivity than through an increase in scale. Farmers in all systems choose to retain agriculture as a base from which to pursue livelihood strategies, especially through off-farm activities. Accounting for this household heterogeneity is important on several levels. First, it provides more granular detail on the process of development in rural areas of Tibet that have been very difficult to access in English literature especially in recent years. Second, the characteristics of the case study in agricultural areas of Tibet — especially the mixed agricultural-pastoral systems, high levels of seasonality, heavy state investment and increased integration with the other regions of China — may contribute to the broader agricultural development study. Third, policymakers in the central and local governments may be interested in how the household heterogeneity may affect — or be affected by — policies to increase agricultural productivity, intensify systems, change the composition of crop-livestock systems, promote rural-urban transformation and pursue environmental objectives, especially grassland degradation.