Many development organizations fund agricultural training for farmers in developing countries.  The purposes of these training programs range from effective use of agricultural inputs (e.g., irrigation water) to cultivation of new crops (e.g., higher profit crops such as fruits) to more effective marketing and post-harvest practices.  Despite the prevalence of such training programs and their frequently large scope, little rigorous research has been conducted on whether training programs targeted at farmers in developing countries are effective. The present study uses a clustered randomized controlled trial to evaluate the effectiveness of a US government-funded farmer training program that trained over 50,000 farmers throughout Armenia.  The overarching objective of the program was to improve farm profitability of smaller scale farms.  The training program had two parts, the first of which focused on region-specific water management techniques to conserve water by emphasizing low-cost irrigation technologies, and the second of which focused on growing new crops or cultivating high-value crop varieties by using higher-quality seeds, establishing greenhouses, or other methods.  Four years after the program began we do not find evidence that training participation substantially improved measures of farmers’ well-being such as household income, poverty avoidance, and consumption. We also investigate impacts of training participation on adoption of new agricultural practices that might suggest that longer-term impacts could develop over time; we do not find evidence of impacts on practice adoption, with the exception that trained farmers are more likely to use safer pesticide practices. The findings from this study suggest that inducing farmers to change their behaviors is challenging, particularly when there are numerous constraints to adopting new practices. The study also brings into focus the possible trade-offs when development organizations seek to have a large programmatic footprint rather than tailoring programs to a narrower population.