Cleavage fracture is one of the most severe failure modes in low alloy ferritic steel structures and the fracture toughness is strongly dependent on the size of the specimen, crack depth and temperature. This thesis investigates how the fracture toughness is affected under conditions that relates to size and constraint effects through a fracture mechanical experiment test series. Furthermore, modeling the experiment with a non-local weakest link model to investigate the importance of the stress measure and how the model parameters can be determined for yielding the best result. The experiment is divided into two sets of testing, a pre-test series to determine an appropriate testing temperature for the main test series. The second set consists of four groups with different sizes and crack depth with 12 specimens in each group. The testing temperature was determined to be -155 ◦C. For the large sized test specimen with a shallow crack, the master curve fails to accurately predict the probability of failure by overestimating the fracture toughness. This is not the case for the small test specimens which indicates that the size effects exceeds the low constraint effects for larger sized test specimens and development of the master curve method is needed. When evaluating the stress state in front of the crack tip in the non-local weakest link model, the results yields the least residual error was obtained when using a large sized specimen with high constraint together with a small sized specimen with low constraint for prediction of the probability of failure.