The aim of this study was to assess the relationship between a number of physiological variables and running velocity at 3 km (v-3km) in a group of male runners. Sixteen well-trained middle- and long-distance runners (mean +/-s: age 22.4 +/- 4.2 years, body mass 63.5 +/- 6.2 kg, VO2 max 73.3 +/- 6.7 ml kg-1 min-1) underwent laboratory treadmill tests to determine their maximum oxygen uptake (VO2 max), running economy at three submaximal velocities (12.9, 14.5 and 16.1 km h-1 or 14.5, 16.1 and 17 km h-1), predicted velocity at VO2 max (v-VO2max), velocity (v-Tlac) and VO2 (VO2-Tlac) at the lactate threshold and their velocity (v-4mM) and VO2 (VO2-4mM) at a blood lactate concentration of 4 mM. Distance running performance was determined by 3 km time-trials on an indoor 200 m track for which the average time was 9.46 +/- 0.74 min. The mean (+/-s) velocities for v-Tlac, v-4mM and v-VO2max were 16.0 +/- 1.8, 17.1 +/- 1.9 and 20.7 +/- 2.1 km h-1 respectively, all significantly different on average (all P < 0.05) from that for v-3km (19.1 +/- 1.5 km h-1). Many of these physiological variables were found to be individually (and significantly at 5%) related to v-3km. The best single predictors of v-3km were v-Tlac and v-4mM (both with a sample correlation, r2 of 0.93), while v-VO2max was slightly poorer (r = 0.86). Neither VO2 max nor running economy was strongly correlated with v-3km. A stepwise multiple-regression analysis revealed that v-Tlac alone was the best single predictor of v-3km and explained 87% of the variability in 3 km running velocity, while the addition of any of the other physiological variables did not significantly improve the prediction of v-3km. We conclude that, in a group of well-trained runners, the running velocity at the lactate threshold was all that was required to explain a large part of the variability in 3 km running performance.