Objective: To measure the frequency of preoperative stress testing and its association with perioperative cardiac events. Background: There is persistent variation in preoperative stress testing across the United States. It remains unclear whether more testing is associated with reduced perioperative cardiac events. Methods: We used the Vizient Clinical Data Base to study patients who underwent 1 of 8 elective major surgical procedures (general, vascular, or oncologic) from 2015 to 2019. We grouped centers into quintiles by frequency of stress test use. We computed a modified revised cardiac risk index (mRCRI) score for included patients. Outcomes included in-hospital major adverse cardiac events (MACEs), myocardial infarction (MI), and cost, which we compared across quintiles of stress test use. Results: We identified 185,612 patients from 133 centers. The mean age was 61.7 (±14.2) years, 47.5% were female, and 79.4% were White. Stress testing was performed in 9.2% of patients undergoing surgery, and varied from 1.7% at lowest quintile centers, to 22.5% at highest quintile centers, despite similar mRCRI comorbidity scores (mRCRI>1: 15.0% vs 15.8%; P=0.068). In-hospital MACE was less frequent among lowest versus highest quintile centers (8.2% vs 9.4%; P<0.001) despite a 13-fold difference in stress test use. Event rates were similar for MI (0.5% vs 0.5%; P=0.737). Mean added cost for stress testing per 1000 patients who underwent surgery was $26,996 at lowest quintile centers versus $357,300 at highest quintile centers. Conclusions: There is substantial variation in preoperative stress testing across the United States despite similar patient risk profiles. Increased testing was not associated with reduced perioperative MACE or MI. These data suggest that more selective stress testing may be an opportunity for cost savings through a reduction of unnecessary tests.