Several methods are used for estimating the anaerobic threshold (AT) during exercise. The aim of the present study was to compare AT values based on blood lactate measurements with those obtained from computerised calculations of different respiratory gas indices. Twelve healthy, well-trained men performed a stepwise incremental test on both treadmill and cycle ergometer. Respiratory gases were measured continuously, and blood samples were drawn every third minute. AT was determined, based on 1) blood lactate concentrations (Lactate-AT), 2) respiratory exchange ratio (RER-AT), 3) V- slope method (Vslope-AT), and 4) ventilatory equivalent for VO₂ (EqO₂-AT). Lactate-AT and RER-AT values showed similar values, both on treadmill and on cycle ergometer. EqO₂-AT showed a trend towards lower values for AT, while Vslope-AT gave significantly lower values for AT for both exercise modes. Bland-Altman plots showed an even distribution of data for RER-AT, while a more scattered and skewed distribution of data was observed when EqO₂-AT and Vslope-AT were compared with Lactate-AT. The study demonstrates that RER-based estimates of AT correlate well with the blood lactate-based AT determination. The RER method is non-invasive and simple to perform, and, in the present study, seemed to be the best respiratory index for estimation of AT.