Background. Whether or not the Australian serviceman is dentally fit for deployment is determined by the dental fitness standard based on NATO’s dental fitness standard for military personnel. Recent operations by Australian, European, and USA defence forces has found that being dentally fit to deployment is not predictive of whether or not a member is likely to have a dental problem in the twelve months between their dental assessments. In order to maintain an effective member in the field, a suitable risk management assessment pre-deployment is essential so that dental casualties are better managed either through having fewer casualties and/or having those casualties better managed in the field. Aim. The aim of this thesis is to establish a Dental Fitness Classification which more accurately predicts a member’s risk of an Unscheduled Dental Visit (UDV) than the existing Denclas system of classification. Methodology. A prospective cohort study was conducted on 875 Australian service men and women in all 3 services in 7 operational bases across Australia with ages ranging from 17-56. Participants were enrolled during their annual dental examination where clinical details were recorded and a questionnaire completed on demographics and lifestyles of participants. UDVs over the next twelve months were enumerated as any visit to the dental centre which did not form part of their planned treatment following an annual dental examination. The data were first analyzed for associations between each putative risk factor and a binary variable indicating whether or not participants had a UDV in the 12 months since enrolment into the study. Variables for the modeling were then selected that yielded moderately significant associations with the odds of a UDV. From the prospective modeling, the parameter estimates from the logistic regression model (including the intercept) were then used to create an Excel spreadsheet for use in dental clinics. The spreadsheet contained the algorithm that calculated a patient’s probability of having a UDV, based on the reduced set of clinical findings and questionnaire responses found to optimize sensitivity and specificity. Results. Prediction models were most accurate when created for the two cohorts; the dentally fit and dentally unfit. The dentally fit model has 8 variables; Denclas 2 (OR=1.50); Unsupported enamel on a an endodontically treated tooth (OR=22.87); Deep periodontal pocketing (OR=13.97); Healthy Diet (OR= 0.07); Oral Hygiene Score (OR=0.27); Years of Service (OR= 3.96 & 3.49); Prior Visits to the Dentist only for the Relief of Pain (OR=0.35) and the interaction term of Denclas and Unsupported enamel on an endodontically treated tooth (OR= 0.09). The model has a sensitivity =82.61% and a specificity =70.23%. The dentally unfit model has 3 variables; Denclas 4(OR=7.08); Large Fillings (OR=7.06) and Toothbrushing times per week (OR=1.95) with a sensitivity =62.86% and specificity = 77.71%. The dentally fit model has excellent discrimination (Area Under the Receiver Operator Characteristic (AUROC) =0.83) and the dentally unfit has a reasonable discrimination (AUROC=0.74) and both are better than the Denclas system in practice in Defence Force Dental Services around the world (OR=1.80, sensitivity= 41.67% and specificity =72.22%). These models can be used to determine a member’s risk of becoming a UDV from a simple chairside tool embedded in a Microsoft Excel® file. Conclusion. There is a set of easily obtainable variables that can more accurately predict an ADF member’s likelihood of becoming a UDV than the present Denclas system which can be adapted to determine a member’s risk at the chairside.