The paper empirically investigates credit access and borrowing costs in Poland's rural financial market. We conduct an econometric analysis based on cross-sectional survey data including formal loans taken in the period 1997-1999. A hedonic regression of the effective interest rate, comprising both the nominal interest rate and additional transaction costs faced by farmers, allows the identification of the determinants of borrowing costs. These determinants can be interpreted as loan attributes and their implicit prices calculated. We proceed in two steps. In the first step, farmers' credit access is estimated by a Probit model. The second step is the hedonic regression, in which the Probit results are taken to test for selectivity. The results support the widely held view that formal lenders tend to discriminate against smaller farms. They also suggest that the presence of devices to screen and signal the quality of borrowers makes borrowing more likely and reduces borrowing costs. Furthermore, the analysis reveals that the choice of the type of bank has a significant effect on borrowing costs. All other loan attributes equal, the traditional institutions for agricultural lending (the cooperative banks and the governmentally controlled Bank for Food Economy) offer between 1.1 and 1.3 percentage point higher effective interest rates as compared with the most favourable terms available, which has implications for a potential future restructuring of the Polish rural banking sector. In addition, there is strong evidence that the government subsidisation of nominal interest rates is severely counteracted by increased transaction costs and an adverse selection of borrowers. However, there is still a net reduction of the effective interest rate by 1.4 percentage point on average, compared to non-subsidised loans. This raises the question whether lending procedures under the government programme are sufficiently streamlined and whether loans are effectively targeted.