Many physicians face increasing stress to see more patients in the same or less time. This leads to crowded appointment schedules and increased schedule disruptions. I examine how physicians respond to schedule disruptions, instrumenting for appointment start time with the office arrival time of the physician’s previous patient. I use novel data from athenahealth, Inc., a national provider of electronic health records, medical billing, and practice management services. I find that when primary care physicians fall behind schedule, they truncate appointment duration, perform fewer in-office procedures, and record fewer diagnoses. The likelihood of a patient revisiting the primary care practice within two weeks significantly increases as a function of delayed appointment start time. Physician ordering behavior also responds to a schedule disruption. In particular, physicians who run behind schedule increase antibiotic and opioid painkiller prescribing and increase referrals of a new patient to a specialist. For patients with preexisting prescription drug regimens, physicians running behind schedule are less likely to change the existing course of treatment. These findings suggest possible unintended consequences of the increasing time pressures placed on physicians by policymakers and private payers. Implications may include higher health care spending and lower quality care. ∗ Correspondence: Department of Health Policy, Harvard University, Cambridge, MA 02138. Tel.: (651) 5877250. Email: email@example.com. Web: http://scholar.harvard.edu/hannahneprash. I am indebted to my committee: Michael Chernew, David Cutler, and Michael McWilliams for input at all stages of this project. I also thank Michael Barnett, Caitlin Carroll, David Chan, Amitabh Chandra, Katherine Donato, Bapu Jena, Tim Layton, Nicole Maestas, Tom McGuire, Ateev Mehrotra, Ellen Montz, Joe Newhouse, Daria Pelech, Alan Rozenshtein, Adam Sacarny, Zirui Song, Teddy Svoronos, Jacob Wallace, and seminar participants for helpful comments and suggestions. I also thank athenahealth, Inc., particularly the research team, for their generous assistance in obtaining the data. I gratefully acknowledge fellowship support through the Agency for Healthcare Research and Quality (AHRQ) trainee program (T32), the National Science Foundation Graduate Research Fellowship, and the AHRQ (IR36H202445-01) Dissertation Grant. All errors are my own.