OBJECTIVES: Demonstrating clinical utility is a challenge for diagnostic companies; MolDx denied 40% of CMS applications due to insufficient clinical utility data.  Trials to prove clinical utility can be expensive and lengthy. Modeling can be a relatively inexpensive method in establishing clinical utility for a novel diagnostic. The objective of this study was to determine the utility of and identify requirements and hurdles for developing clinical utility models.    METHODS: We conducted a qualitative review of 15 clinical utility models for novel diagnostics across diverse therapeutic areas and encompassing screening, diagnostic and monitoring tests.  Models were assessed based on data requirements, validity of outcomes, and ability to secure reimbursement. RESULTS: Clinical utility depends on: (1) test performance (relative to standard of care), (2) physician practice change (confidence in test results), (3) patient compliance/behavioral change, (4) availability and proven benefits of alternative treatment course, (5) and/or reductions in adverse events.   Modeling for clinical utility is most effective in areas where clear treatment protocols exist and evidence supporting the efficacy of an alternative treatment is robust. Evidence supporting practice change and patient compliance are frequently unknown, but may be informed by literature, claims analysis and/or EMR data.  Modeling clinical utility is most challenging when treatment guidelines are broad and outcomes evidence is not well differentiated.   CONCLUSIONS: Modeling is a logical first step in assessing clinical utility.  It's a cost-effective way of identifying a target population and best test placement in the continuum of care through comparison of alternative strategies.  Additionally, it can be useful in identifying evidence gaps and prioritizing data collection.  However unless the data supporting the model is strong, it is insufficient on its own to secure reimbursement.   Analysis of claims and EMR data can be an excellent data source for supporting diagnostic utility modeling.