OBJECTIVES: All people with type 2 diabetes should receive dietary advice.  Some patients may benefit from additional dietary counseling.  This study used recently developed statistical methods to estimate the efficiency frontier for individualized allocation of dietary counseling services.  I.e., for each level of total health care expenditure, we estimate the individualized allocation of services that maximizes clinical benefit in the population.   METHODS: People newly diagnosed with type 2 diabetes were identified retrospectively from electronic health records and classified as receiving vs. not receiving dietary counseling.  An individualized effectiveness score for achievement of HbA1c < 7% with vs. without counseling was estimated using multivariable logistic regression.  Demographics, vitals, comorbidities and lab values served as candidate predictors. Prediction models were validated in a held-out sample.  The efficiency frontier was estimated as the convex hull of the set of HbA1c control rates and total costs achievable by allocation of dietary counseling based on the effectiveness score.  RESULTS: Among 11,819 patients newly diagnosed with type 2 diabetes, 22% received dietary counseling and 74% achieved HbA1c control.  Greater HbA1c, body mass index and age at the time of diagnosis were associated with greater effectiveness of dietary counseling.  Allocation of all newly diagnosed diabetics to dietary counseling was estimated to increase the rate of HbA1c control to 80% at a cost of 56 USD per patient vs. current practice.  An efficient allocation rule (counseling only the 55% of patients predicted to benefit most) achieved the same 80% control rate with an incremental cost of 6 USD per patient.   CONCLUSIONS: Retrospective analysis of real-world data identified opportunities to improve diabetes outcomes vs. current practice with minimal expense through individualized allocation of dietary counseling.  This result warrants validation in separate data.  The analytical methods warrant broader use to investigate efficient allocation of treatments.