There is general agreement that the oxygen demand of level running is similar for both the treadmill (TM) and overground situations at speeds under 260 m X min-1. However, controversy exists with regard to inclined running. The prevailing view, represented by the ACSM prediction formulas, is that overground hill running is theoretically more costly than inclined treadmill running. This study was designed to investigate the problem from an empirical standpoint. Seven male subjects performed overground and TM running at two grades (0 and 5.7%) over a range of speeds between 136-286 m X min-1. For the outdoor trials, subjects covered a distance of 950 m at a constant pace, and expired gas was collected over the last 150 m. Matching trials were then performed on the treadmill at the same speed and % grade. Regression lines were calculated for speed vs oxygen consumption (VO2). For TM and overground level running, these were: VO2 (ml.kg-1.min-1)= 0.222 X speed (m.min-1) - 1.33 and VO2 (ml.kg-1.min-1) = 0.202 X speed (m.min-1) + 3.21 respectively. The regression lines from TM and overground inclined running were: VO2 (ml.kg-1.min-1) = 0.237 X speed (m.min-1) + 7.53. and VO2 (ml.kg-1.min-1) = 0.233 X speed (m.min-1) + 7.78 respectively. A 2 X 3 X 2 ANOVA revealed that the differences between mean values for VO2 for level TM running vs level overground running and grade TM running vs grade overground running were not statistically significant (0.10 less than P less than 0.25).(ABSTRACT TRUNCATED AT 250 WORDS)