Calving difficulty, biologically and statistically, have special characteristics which require researchers to use more complex models for its analysis. First, it is influenced by maternal effect and therefore the models should take this aspect into account. This is why animal models with maternal effect or sire-maternal grandsire models are employed. Secondly, it is recorded subjectively using usually five scores, one for a normal delivery to five for a Caesarean parturition. In other words, calving difficulty is a discrete trait and is not distributed normally. In theory, methods for analyzing continuous data are not appropriate for categorical data. In the 1980s, researchers developed a standard threshold model (a nonlinear model) for genetic analysis of these traits (Harville et. al. 1984). In the threshold model each phenotype in the categorical scale is associated with an unobservable underlying continuous variable with normal distribution. The objective of this study was to estimate (co)variance components of calving difficulty and genetic parameters using a threshold model.