Gait analysis is used for the assessment of walking ability of children with cerebral palsy (CP), to inform clinical decision making and to quantify changes after treatment. To simplify gait analysis interpretation and to quantify deviations from normality, some quantitative synthetic descriptors were developed over the years, such as the Movement Analysis Profile (MAP) and the Linear Fit Method (LFM), but their interpretation is not always straightforward. The aims of this work were to: (i) study gait changes, by means of synthetic descriptors, in children with CP that underwent Single Event Multilevel Surgery; (ii) compare the MAP and the LFM on these patients; (iii) design a new index that may overcome the limitations of the previous methods, i.e. the lack of information about the direction of deviation or its source. Gait analysis exams of 10 children with CP, pre- and post-surgery, were collected and MAP and LFM were computed. A new index was designed as a modified version of the MAP by separating out changes in offset (named OC-MAP). MAP documented an improvement in the gait pattern after surgery. The highest effect was observed for the knee flexion/extension angle. However, a worsening was observed as an increase in anterior pelvic tilt. An important source of gait deviation was recognized in the offset between observed tracks and reference. OC-MAP allowed the assessment of the offset component versus the shape component of deviation. LFM provided results similar to OC-MAP offset analysis but could not be considered reliable due to intrinsic limitations. As offset in gait features played an important role in gait deviation, OC-MAP synthetic analysis was proposed as a novel approach to a meaningful parameterisation of global deviations in gait patterns of subjects with CP and gait changes after treatment.