Food-for-work (FFW) programs are commonly used both for short-term relief and long-term development purposes. In the latter capacity, they are increasingly used for natural resources management projects. Barrett, Holden and Clay (forthcoming) assess the suitability of FFW programs as insurance to cushion the poor against short-term, adverse shocks that could, in the absence of a safety net, have permanent repercussions. In this paper we explore the complementary question of FFW programs' potential to reduce poverty and promote sustainable land use in the longer run through induced changes in investment patterns. FFW programs commonly aim to produce or maintain potentially valuable public goods necessary to stimulate productivity and thus income growth. Among the most common projects are road building, reforestation, and the installation of terracing or irrigation. In the abstract, public goods such as these are unambiguously good. There is a danger, however, that such programs could discourage private soil and water conservation and crowd out private investment. How important are such effects and when are these effects small or large and when and how can they be reduced? How do market characteristics, timing and design of FFW programs affect this? When, where and how can FFW programs more efficiently reduce poverty and promote more sustainable land management? The paper aims to answer these questions. Much recent empirical research has focused on the shorter-term targeting issue of whether FFW and related workfare programs efficiently target the poor (Dev 1995, Von Braun 1995, Webb 1995, Subbarao 1997, Clay et al. 1998, Devereux 1999, Jayne et al. 1999, Ravallion 1999, Teklu and Asefa 1999, Atwood et al. 2000, Gebremedhin and Swinton 2000, Haddad and Adato 2001, Jalan and Ravallion 2001). Much less research has been focused on the longer-term effects of FFW. Yet the large share of hunger worldwide arises due to chronic deprivation and vulnerability, not short-term shocks (Speth 1993, Barrett 2002). Also most of the FFW programs in Ethiopia have long-term development goals and are formally distinguished from the disaster relief FFW programs (Aas and Mellemstrand 2002). It is therefore appropriate to evaluate these programs based on their long-term goals and not only on the basis of short-term targeting. In a case study in Tigray Aas and Mellemstrand (2002) found that the FFW recipients considered the long-term benefits of FFW as more important than the short-term benefits of food provision. FFW programs may produce valuable public goods. For example, Von Braun et al. (1999) report multiplier effects of a FFW-built road in the Ethiopian lowlands. Public provision of public goods may be socially desirable because private investment in soil and water conservation and tree planting may be well below socially optimal levels due to poverty and market imperfections (Holden, Shiferaw and Wik 1998, Holden and Shiferaw 2002, Holden and Yohannes 2002, Pender and Kerr 1998), tenure insecurity (Gebremedhin and Swinton 2000, Holden, Benin, Shiferaw and Pender 2003), lack of technical knowledge and coordination problems across farms (Hagos and Holden 2002). There is, however, also a danger that FFW programs crowd out private investments (Gebremedhin and Swinton 2000). We analyze these issues using multiple methods. First, section II introduces a simple theoretical framework for understanding the analytically ambiguous effects of FFW programs on sustainable land use patterns. We first present the basic intuition in a static framework to illustrate the selection, crowding out and targeting issues, before generalizing it to a dynamic model to illustrate the possible insurance and crowding in effects of FFW. Section III then uses an applied, dynamic bio-economic farm household model applied to a less-favoured area in Ethiopia to investigate via numerical simulation how household welfare and land use patterns vary with changes in environmental and FFW program design parameters. Section IV presents econometric evidence based on survey panel data from northern Ethiopia to assess the relationship between FFW and private investment in conservation. Section V discusses our findings and fleshes them out a bit with further empirical evidence. Section VI concludes.