Ductile fracture is a common limit state of frameworks of steel structures and can be predicted by incorporating models for fracture initiation and propagation in finite element (FE) analyses. However, accurate prediction of fracture relies on unique fracture parameters for a given material, requiring precise predictions of fracture strain and stress states obtained through coupon tests and accompanying FE data analysis. In recent decades, various methods have been proposed to predict ductile fracture initiation, including various design geometries of coupons, test setups and FE analyses, which may lead to inconsistent fracture strains and stress states. Hence, there is a need for proposing a standard procedure for the design, test and accompanying FE-based calibration of fracture parameters, as pursued in this paper. Given that ductile fracture initiates under significant plastic strain, a constitutive model for the full strain range is required, as also proposed in this paper. Moreover, to reduce the experimental and data analytical efforts, an empirical method is proposed that uses only a small number of coupon tests to calibrate the fracture parameters. The method encompasses three levels wherein one, two, or three coupon tests are required. All in all, the paper presents a methodology for determining the full-range true stress-strain curve and the initiation of fracture, including a new post-necking constitutive model and methods for determining the free parameters of the post-necking and fracture initiation models. While applied to steel in this paper, the proposed methodology is potentially also applicable to other metals.