The influence of low and high carbohydrate diets on the relationship between blood lactate concentration ([Lac]) and work load (WL) in incremental exercise tests (cycle ergometer) and endurance tests was evaluated in trained subjects. The relationship between relative work load (WLrel) and [Lac] in arterialized blood was compared in untrained subjects (UT) and trained male athletes (TR) after 2 days without training while consuming a high carbohydrate diet (HCD). In both groups [Lac] of 2 mmol.l-1 was reached at about 60% [(mean +/- SD) UT 57.7% +/- 6%, TR 62.7% +/- 3.8%] and 4 mmol.l-1 at about 75% (UT 75.2% +/- 3.6%, TR 77.8 +/- 2.2) of the maximal work load (WLmax). In eight cyclists the relationship between [Lac] and WL was not influenced by a 13-day training camp; however, heart rate was lower after the training camp. During their normal training programme, trained subjects had high relative work loads at their [Lac] thresholds, but after an HCD combined with an interruption of the training of 3 days, the relationship between [Lac] and WLrel was the same as in UT. In six TR a low carbohydrate diet (LCD) combined with training led to high absolute (WLabs) and WLrel at [Lac] at 2 and 4 mmol.l-1; an HCD combined with 3 days without training led to low WLabs and WLrel at the same [Lac] and to higher WLmax. In spite of the apparently lower endurance capacities TR were able to work significantly longer after HCD than after LCD (23 +/- 10.5 min and 49 +/- 16.2 min, respectively) at 65% of their WLmax. The variability of the relationship between [Lac] and WL following the dietary regimes leads to the conclusion that the "typical" [Lac] versus WL curve of endurance TR may result from a permanent glycogen deficiency.