Background Coronary artery calcium scoring (CACS) improves management of chest pain patients. However, it is unknown whether the benefit of CACS is dependent on the clinical likelihood (CL). Objectives This study aims to investigate for which patients CACS has the greatest benefit when added to a CL model. Methods Based on data from a clinical database, the CL of obstructive coronary artery disease (CAD) was calculated for 39,837 patients referred for cardiac imaging due to symptoms suggestive of obstructive CAD. Patients were categorized according to the risk factor–weighted (RF-CL) model (very low, ≤5%; low, >5 to ≤15%; moderate >15 to ≤50%; high, >50%). CL was then recalculated incorporating the CACS result (CACS-CL). Reclassification rates and the number needed to test with CACS to reclassify patients were calculated and validated in 3 independent cohorts (n = 9,635). Results In total, 15,358 (39%) patients were down- or upclassified after including CACS. Reclassification rates were 8%, 75%, 53%, and 30% in the very low, low, moderate, and high RF-CL categories, respectively. Reclassification to very low CACS-CL occurred in 48% of reclassified patients. The number needed to test to reclassify 1 patient from low RF-CL to very low CACS-CL was 2.1 with consistency across age, sex, and cohorts. CACS-CL correlated better to obstructive CAD prevalence than RF-CL. Conclusions Added to an RF-CL model for obstructive CAD, CACS identifies more patients unlikely to benefit from further testing. The number needed to test with CACS to reclassify patients depends on the pretest RF-CL and is lowest in patients with low (>5% to ≤15%) likelihood of CAD.