
Compaction quality control in pavement construction is commonly governed by statistical acceptance specifications intended to account for construction and testing variability. However, these approaches do not explicitly quantify the reliability of the lot mean and may become increasingly conservative as the number of quality-control tests increases. This issue is increasingly important as intelligent compaction technologies enable the efficient collection of large datasets of lot quality. To address these limitations, a reliability-based acceptance rule is proposed in which acceptance is directly governed by the probability that the true lot mean (μtrue) is greater than the specified requirement. A new reliability multiplier (kR), defined in terms of a target lot failure probability (Pf∗), which represents the maximum allowable probability that μtrue is below the limit (L), enables acceptance decisions to be expressed using a single, intuitive reliability measure that is readily communicated among stakeholders. The framework is demonstrated using nuclear density gauge measurements from three test sites. Results show stable, monotonic acceptance behaviour with increasing sample size, while the computed lot failure probability decreases systematically as the uncertainty in the estimated lot mean decreases. Sensitivity analyses further indicate that more stringent reliability targets require greater testing effort before acceptance can be achieved. A practical implementation workflow is also proposed, including guidance for reliability selection, representative sampling, and treatment of marginal lots. Overall, the proposed framework provides a transparent and practical approach for reliability-informed compaction quality control.