The study tried to identify the determinants influencing Commercial banks decision to ration agricultural credit in South-Western, Nigeria. Data for the analysis were sourced from the agricultural credit transactions of the banks. Evidence from the multinomial logit model estimated shows that the borrowers are heterogeneous. Farm size, previous income, enterprise type, coop membership, household net-worth and agricultural commercialization level are positive and significantly associated with the classification of the two groups relative to the reference group. The significant variables affect both the probability of classification and the utility of the banks in their decision making. The partial elasticities of farm size are elastic at 1.5380, 1.2796, and 1.0065 for the groups as classified. The quasi elasticities for the household net-worth and agricultural commercialization variables are all 

elastic for all the groups. The quasi–elasticity for the income variable for the first group is elastic at 1.4278 and for the second group at 1.2551. This variable is inelastic for the reference group. It is recommended that Banks borrowing decisions must be group specific and not general. There is also the need to find an innovative way of meeting the need of the rejected group in terms of Micro finance arrangements. The blanket policy approach will not lead to the desired results of easy access to agricultural production credit by the resource poor farm households in Nigeria.