ABSTRACT : Implementing an on-farm storage system for agricultural products becomes a complex decision problem when considering risks, benefits, uncertainties and alternatives. This study aimed at applying the multicriteria decision aid (MCDA) approach to select the infrastructure to be implemented, taking into account variables and farmer’s goals. As a result, we concluded that the adopted approach was useful since it allowed a better understanding of the problem and provided decision-making support. KEYWORDS : grain storage, investment, multicriteria decision aid. INTRODUCTION Implementing an on-farm grain storage system is a complex decision problem because it covers several variables of significant uncertainty level, such as price fluctuations, freight costs, logistics, opportunity cost, strategies of utilization for a storage system and interest rate. Additionally, “grain production in Brazil has increased at an astounding rate, culminating in record harvests”, but “a few Brazilian regions show a critical situation regarding their storage systems” (AGUIAR, 2013), as well as transport and shipping deficiencies (MACHADO, 2013). It is noteworthy that there is no consensus on the static storage capacity required for the country, and several authors support the FAO’s recommendation (Food and Agriculture Organization), which is 120% of the annual grain production in the country (GALLARDO et al., 2010). In the case of soybean, Brazil barely hold stocks and currently “almost all soybean produced is exported or consumed” (CONAB, 2015). Thus, it is of major importance to consider specific characteristics concerning grain production and transport in Brazil, while assessing a storage system implementation. The first aspect to consider is the “soybean market and the factors that can impact on supply and demand” (BARBOSA & VIEIRA, 2013). Brazil has a strong exporting agriculture and its main commodity is soybeans. However, its storage seems to be commercially not interesting because “in Brazil and in the other countries in the Southern Hemisphere harvest extends from March to May, while in the Northern Hemisphere it takes place in September/October (in the United States)”. Thus, “Brazilian preferential export period” occurs from “May to August, i.e., prior to American crop harvest” (FERREIRA et al., 1993). In addition to occurring during North American off-season, seasonal price changes preclude Brazilian harvest storage (TOMAZELA, 2010), and might become a risk due to their high volatility. Therefore, it is important to consider additional benefits, or “utility gains” (“convenience yield”), warranting storage (POYNDER, 1999). Another important variable is interest rate. In addition to stand for “opportunity cost of product storing” (FERREIRA et al., 1993), if too high, it can make impractical to finance new static capacity expansion projects (WEBBER, 2011), or even derail the operation of existing warehouses, as occurred with public structures that have been “disabled by activity change or scrapping” during the 1990s (FERNANDES & ROSALEM, 2014). In this context, it is highlighted that the return on