Major forming limit prediction models and calibration methods are reviewed briefly and their advantages and disadvantages are discussed. Two modified Marciniak-Kuczynski (M-K) models and one modified NADDRG (Keeler-Brazier) model are also presented which have some advantages over conventional models. In the first modified M-K model, material non-homogeneity has been substituted for geometrical non-homogeneity to reduce the sensitivity of the traditional model to variations of the initial non-homogeneity. Using this important advantage, a semi-empirical relation is proposed to predict the value of the initial material non-homogeneity. In the second modified M-K model, the conventional calibration method (which requires an experimental point, corresponding to plain strain condition, to find the initial non-homogeneity and calibrate the model) is revised and the uniaxial tensile point, which is easily obtained, is proposed to be used in the calibration process. Finally, a modified NADDRG relation is suggested which predicts the Forming Limit Curve (FLC) much more accurately and for various types of materials (i.e. not only low carbon steel). This new relation uses the value of strain rate sensitivity index. The results obtained from these models and calibration methods are compared with experimental results. The proposed modifications are very easy to implement; could be applied to various materials; require few and easily measurable parameters and are in good agreement with experiments.