Liquid biofuels are considered as a strategy for achieving energy security, stimulating rural development and mitigating climate change. Low-income countries largely focus on biofuel production from non-food crops and trees grown on underutilized lands, in order to minimize food, fuel and environmental trade-offs. Unfounded claims on the potential of such alternative biofuel crops led to large-scale investments, especially in jatropha (Jatropha curcas) monoculture plantations. Productivity and economic outcomes proved to be poor, however, which subsequently led to a global downturn in research and investment, leaving the potential of other species and approaches unexplored. In Chapter 1, we hypothesize that agroforestry-based approaches that carefully integrate a mix of native biofuel tree species into the existing farming system, offer unique opportunities in comparison with jatropha monocultures, including limited risk, increased by-product valorization, and expanded ecosystem services provision. This PhD study aims at evaluating the bio-economic potential of agroforestry-based biofuel systems by quantifying three critical success factors: farmer perception and adoption, oilseed yield, and economic impact. The in-depth empirical analysis is based on a mature agroforestry-based biofuel program in Hassan district, South India. Since 2007 this program stimulates the cultivation of native oilseed tree mixtures on farms through a range of extension and implementation activities. In Chapter 2 we assess farmer perception and adoption of oilseed trees and the biofuel value chain in Hassan district. To this end, cross-sectional survey data of 396 farm-households in Hassan district are collected. We find that although 60% of the farmers cultivate oilseed trees, oilseed collection rates are generally low (13%), as are oil expeller use (< /1%) and biofuel marketing (< /1%). To examine the impact of the biofuel program on adoption, we use regression analyses addressing various forms of selection bias. We find that various activities of the biofuel program stimulate oilseed tree cultivation but not oilseed collection. Low seed prices, high opportunity costs of labour, and value chain underdevelopment are major factors impeding households to collect seeds from planted or wild oilseed species. In Chapter 3 we use a labelled choice experiment to assess the same farmers' preferences for alternative production systems, value chain organisations and market developments. This allows predicting the extent to which hypothetical changes in these characteristics could change the likelihood of adoption. Our results demonstrate that biofuel programs can benefit from ex ante analyses to improve their design. We find that most farmers (71%) are likely to adopt biofuel trees in most scenarios, especially species with relatively high yields, low labour requirements and high oilseed prices. Nevertheless, value chain reorganization through contracting and labour provision proves to be the key lever to stimulate adoption. This calls for further research on effective contract design and implementation, and for developing alternative business models. Chapters 2 and 3 indicate that pongamia (Millettia pinnata) is the species with highest adoption potential. However, yield levels and dynamics of this undomesticated species remain poorly understood, despite of the fundamental role they play for its biofuel potential. In Chapter 4 we address the critical lack of scientific evidence by collecting primary seed and oil yield data from 81 pongamia trees in South India, and explore which factors might determine yields using empirical models. Our results indicate that annual seed and oil yields for pongamia trees vary widely, but generally remain below 2500 kg/ha and 1000 liter/ha, respectively. This current field performance is substantially lower than commonly reported figures in the literature. Furthermore, our results suggest that a complex interplay between genotype, environment and agronomy leads to large spatiotemporal variation in yields, and that this interplay remains poorly understood. Long-term yield monitoring is required to get better insights into yield mechanisms, and to assess the actual potential of pongamia as a reliable and significant source of biofuel feedstock. In Chapter 5 we address common methodological shortcomings in the literature on profitability of novel biofuels, by developing a sound framework for quantifying the long-term financial performance of agroforestry-based biofuel value chains. The framework is applied to calculate profitability of pongamia cultivation and processing in Hassan district. The results show that pongamia cultivation has limited financial potential, and is only profitable in small-scale settings, in the middle to long term and for a subset of farmers. If biodiesel is envisaged as the end product, the value chain requires substantial fiscal and marketing support to be economically viable. For current prices, financial performance is much higher if the seed oil is marketed instead of processed to biodiesel. These findings are case-specific, while the developed framework opens the door to comprehensive investigation of the financial performance of other oilseed tree species and in other regions. The interdisciplinary framework allows concluding that the potential of agroforestry-based biofuel systems as sources of energy, income and employment is currently very limited. The program in South India succeeds as an agroforestry program but not as a biofuel program. Low profitability impedes local farmers to collect oilseeds, while processors have no financial incentive to convert oil to biodiesel. We find that similar challenges pertain to small-scale agroforestry systems as to jatropha-based plantation systems, although the former are a Low-Risk High-Diversity approach to build feedstock for the future. The wider validity of our findings should be further explored to determine in which niches alternative biofuel crops may still have potential.