Rationale Missing data in routine practice settings in primary care present a major issue in determining the effectiveness of clients' outcomes in primary care. This is particularly true for client self-report measures, one of which is the CORE-OM which is widely used in primary care settings. In order to build a robust evidence base in practice settings, user-friendly procedures need to be disseminated to practitioners. Aim To provide a statistically sound yet simple method for calculating a predicted CORE-OM end-of-therapy score for those clients who did not complete an outcome measure at discharge. Results The models created were statistically able to predict outcome scores, using clients' pre-therapy CORE-OM score and therapist-identified problems. When testing the models on two independent datasets, there was some variability with predictive ability. Where the significance level for the difference between predicted and actual score was problematic, this was associated with small effect sizes. Conclusion It is possible to construct equations for men and women which, with an inevitable degree of error, can be used to predict reasonably well CORE-OM scores for those patients in primary care who did not complete an outcome measure at discharge.