Allelic drop-out continues to be a challenge to forensic geneticists when interpreting crime scene evidence and calculating the evidential weight. Methods exist for estimating the probability of allelic drop-out, where the most promising methods use the signal intensities as input for quantifying the drop-out probability. Using data from real crime cases, we demonstrate that taking degradation of the biological material and truncation of the data due to a detection threshold into account is superior to previous approaches. An additional correction for stuttering effects showed limited improvement.