There are a number of increasingly sophisticated techniques available for orthognathic treatment planning. All are based on the determination of the skeletal pattern and the position of the dentition. However, they all suffer from difficulties associated with predicting the soft tissue profile. The aim of this retrospective cephalometric investigation was, therefore, to compare the ability to predict accurately the outcome of orthognathic treatment using the 'hand planning' technique and the orthognathic planning and analysis (OPAL) computer program, with an emphasis on the soft tissue profile. Seventy adult subjects were divided into two groups not specific for gender or age: the Class III patients had undergone bimaxillary surgery and the Class II patients sagittal split mandibular advancement. In each group, the pre-treatment and post-debond lateral cephalograms were utilized to calculate the actual orthodontic and surgical movements. These values were then used to produce a prediction using both the hand planning technique and the OPAL program. The resultant predictions were digitized using a customized computer program and compared with the actual outcome. The results show that there was marked individual variation when planning by hand and using the OPAL program. In the mandibular surgery group, hand planning and OPAL were of similar accuracy and few points differed significantly between prediction and outcome. However, for the bimaxillary group, a number of points showed bias and the hand planning technique appeared to be more accurate than the OPAL program, particularly in the region of the lips. Although the usefulness of predictions is acknowledged, these results suggest that they should be used with a certain amount of caution.