Background Metabolic outcomes in type 1 diabetes remain suboptimal. Disease modifying therapy to prevent ?-cell loss presents an alternative treatment framework but the effect on metabolic outcomes is unclear. We, therefore, aimed to define the relationship between insulin C-peptide as a marker of ?-cell function and metabolic outcomes in new-onset type 1 diabetes. Methods 21 trials of disease-modifying interventions within 100 days of type 1 diabetes diagnosis comprising 1315 adults (ie, those 18 years and older) and 1396 children (ie, those younger than 18 years) were combined. Endpoints assessed were stimulated area under the curve C-peptide, HbA1c, insulin use, hypoglycaemic events, and composite scores (such as insulin dose-adjusted A1c, total daily insulin, U/kg per day, and BETA-2 score). Positive studies were defined as those meeting their primary endpoint. Differences in outcomes between active and control groups were assessed using the Wilcoxon rank test. Findings 6 months after treatment, a 24·8% greater C-peptide preservation in positive studies was associated with a 0·55% lower HbA1c (p<0·0001), with differences being detectable as early as 3 months. Cross-sectional analysis, combining positive and negative studies, was consistent with this proportionality: a 55% improvement in C-peptide preservation was associated with 0·64% lower HbA1c (p<0·0001). Higher initial C-peptide levels and greater preservation were associated with greater improvement in HbA1c. For HbA1c, IDAAC, and BETA-2 score, sample size predictions indicated that 2–3 times as many participants per group would be required to show a difference at 6 months as compared with C-peptide. Detecting a reduction in hypoglycaemia was affected by reporting methods. Interpretation Interventions that preserve ?-cell function are effective at improving metabolic outcomes in new-onset type 1 diabetes, confirming their potential as adjuncts to insulin. We have shown that improvements in HbA1c are directly proportional to the degree of C-peptide preservation, quantifying this relationship, and supporting the use of C-peptides as a surrogate endpoint in clinical trials.