Objective: To assess reliability of scoring plaster models and their 3D digital copy of children with complete unilateral cleft lip and palate (CUCLP) using a continuous scale (10-cm visual analog scale [VAS]) and a categorical scale (GOSLON Yardstick). Design: Reliability observational study involving 3 trained GOSLON Yardstick assessors blinded to the origin of the models. Patients: Models from 35 New Zealand (NZ) and 35 Oslo CUCLP patients were standardized and randomly ordered before rating. Outcome Measures: Assessments were undertaken using the GOSLON and the VAS for both model formats. Twenty percent of sample were randomly selected and rescored at the first assessment, and whole sample was rescored 1 week later. Weighted κ was used to assess GOSLON reliability, while correlation was used for the VAS. Results: The VAS and GOSLON intra- and inter-rater agreement was similar for both model formats. Repeat measurements on the day have similar intra-rater reliability as repeat measurements at least a week part. There was no significant difference between the 2 model formats, and both the GOSLON and VAS found the NZ models were significantly worse than Oslo. Conclusions: A 10-cm VAS is a reliable method to assess dental arch relationships and appears to have good face validity when compared to GOSLON. The VAS allows for statistically robust rankings of the dental arch relationships, although more studies will be required to enable the VAS scores to have greater clinical meaning. The 3D digital models can be used for GOSLON and VAS rankings with a high degree of reliability.