Study hypothesis: Physician service time varies with patient service category, length of stay, and intensity of service. Design: Prospective time study of emergency physician services. Physicians recorded the beginning and ending times of each service episode offered to a patient (whether at the bed-side or occurring elsewhere in the department). Each episode was defined as an “interaction,” with the total service time offered to a patient being the sum of all interactions for that patient. Length of stay was the time interval from when the patient registered in the emergency department to when the patient was released. Intensity of service was calculated as service time divided by length of stay. Setting: University-affiliated community teaching hospital. Type of participants: One thousand three hundred forty-seven ED patients were entered into the study for nonselected (514), walk-in (637), observation (52), laceration repair (102), or critical care (42) services. Six of 12 physicians in the group staffing the ED participated in the study. Patient data were entered onto study cards when the service was offered. Patients were entered into the study consecutively except when the physician became too busy to see one patient at a time and accurately enter time data; such interruptions occurred for 18% of the patients. Results: Physician service time for nonselected service patients (24.2 minutes per patient; 95% CI, 23.1–25.3) was consistent with ACEP's findings for nonselected services offered by emergency physicians (22 minutes per patient). Physician service time did not vary significantly from the standard for laceration repair patients (25.0 minutes per patient; 95% CI, 22.6–27.4) but did vary significantly from the standard for walk-in (9.8 minutes per patient; 95% CI, 9.3–10.3; P < .05), observation (55.6 minutes per patient; 95% CI, 50.7–60.5; P < .05), and critical care patients (31.9 minutes per patient; 95% CI, 26.2–37.6; P < .05). Walk-in and laceration repair patients had a single physician-patient interaction (1.3 per patient and 1.1 per patient, respectively), consistent with a discrete service offered during episodic care. Observation and critical care patients had multiple physician-patient interactions (6.3 per patient and 2.6 per patient, respectively) over an extended period, which is consistent with additional services being offered during their period of observation/holding. Conclusion: Case mix of patient services affects emergency physician workload and should be considered in planning departmental staffing needs.