The dimension and determinants of poverty as well as the level of persistence in poverty among 540 small scale farmers in central Nigeria was investigated in this study. It was found that 88% of the farmers were classified as poor based on average nominal income of NGN 1 584,267.56 while 92% were classified poor based on per capita income of NGN98,444.26. The predicted poverty categories (PCs) of the respondents from MNL were 74% extremely poor, 21% moderately poor and 5% non-poor while DOGIT predicted 69%, 24% and 6% respectively which are quite different from the 67%, 25% and 8% respectively as specified by the data in contrast. These results indicated that the ordering of the respondents in the poor category is sensitive to the model employed and that poverty among the farmers is more deep-seated than the data suggest. There was no convergence in the estimation of the level of captivity which may be an indicator that captivity among the respondents was absolute indicating the possibility that the farmers are reluctant to imbibe poverty liberation technologies. Among the key policy variables affecting the probability of poverty among the farmers are their states of domicile, those in Mariga LGA 2 as against those from other LGAs, number of other dependents, the amount of agrochemicals applied to the farm plots, the quantity of planting materials, total size of farm plots, membership of farmers organisations, the major occupation, the amount of owner capital invested in the farm business and the amount of credit obtained in the farming season under investigation. However, their influence are not uniform across PCs, while some of the variables tended to accentuate the probability of the farmers being poor, others de-accentuate the probability. From the foregoing, strong persuasion is needed to make the farmers exit poverty and further poverty reduction strategies must be done by targeting the factors that accentuate the probability of the farmers remaining poor.