The aim of this study was to determine exercise intensity and metabolic response during singles tennis play. Techniques for assessment of exercise intensity were studied on-court and in the laboratory. The on-court study required eight State-level tennis players to complete a competitive singles tennis match. During the laboratory study, a separate group of seven male subjects performed an intermittent and a continuous treadmill run. During tennis play, heart rate (HR) and relative exercise intensity (72 +/- 1.9% VO2max; estimated from measurement of heart rate) remained constant (83.4 +/- 0.9% HRmax; mean +/- s(x)) after the second change of end. The peak value for estimated play intensity (1.25 +/- 0.11 steps x s(-1); from video analysis) occurred after the fourth change of end (P< 0.005). Plasma lactate concentration, measured at rest and at the change of ends, increased 175% from 2.13 +/- 0.32 mmol x l(-1) at rest to a peak 5.86 +/- 1.33 mmol x l(-1) after the sixth change of end (P < 0.001). A linear regression model, which included significant terms for %HRmax (P< 0.001), estimated play intensity (P < 0.001) and subject (P < 0.00), as well as a %HRmax subject interaction (P < 0.05), accounted for 82% of the variation in plasma lactate concentration. During intermittent laboratory treadmill running, % VO2peak estimated from heart rate was 17% higher than the value derived from the measured VO2 (79.7 +/- 2.2% and 69.0 +/- 2.5% VO2peak respectively; P< 0.001). The %VO2peak was estimated with reasonable accuracy during continuous treadmill running (5% error). We conclude that changes in exercise intensity based on measurements of heart rate and a time-motion analysis of court movement patterns explain the variation in lactate concentration observed during singles tennis, and that measuring heart rate during play, in association with preliminary fitness tests to estimate VO2, will overestimate the aerobic response.