Although five minutes is the sampling interval mentioned by the American Society of Anesthesiologists for monitoring blood pressure and heart rate during anesthesia, most patients are monitored more closely by continuous auscultation and with the help of automated instruments. Yet this difference between the interval recommended and that actually used indicates that sampling intervals are not defined clearly enough. Therefore, we present three methods with which to determine sampling intervals during monitoring. To explore the feasibility of these methods we examined data gathered every 7.5 seconds during three typical, noncatastrophic physiologic perturbations induced in an anesthetized dog. We chose hypercapnia secondary to rebreathing, hypotension secondary to deep anesthesia, and hypoxemia secondary to a low concentration of inspired oxygen as realistic examples of what can occur during operation and anesthesia. We studied three variables: respired carbon dioxide, femoral arterial blood pressure, and oxygen saturation of hemoglobin (pulse oximeter). The data obtained during monitoring were subjected to three methods of analysis: (1) recording of sets of data, with various starting times, at five-minute intervals only (moving grid); (2) Fourier analysis; and (3) analysis of slopes. For the data of the experiment, the Fourier analysis yielded, on average, longer sampling intervals than did the analysis of slopes.