Using recent advances in the stability analysis of current ripples, a new model for the calculation of duration of sediment deposition from decelerating high-density turbidity currents is proposed. The model, named TDURE, refines a duration model proposed by Allen (1991) (J. Sediment. Petrol. 61, 291-295) by calculating the accumulation time of the rippled T(c)-division separately from the massive and plane parallel-laminated T(ab)-divisions in Bouma-type turbidites. TDURE consists of three modules. In the first module, the accumulation time for the T(a)- and T(b)-divisions is approximated by assuming a linear decrease in sedimentation rate with height in the turbidite. As in the original model, a straight line is fitted through inferred sedimentation rates at the T(ab)- and T(bc)-boundaries. In the second module, angle of climb of ripples, thickness of the T(c)-division and grain size distribution are used in empirical relationships between ripple migration rate and a grain-related mobility parameter to estimate the accumulation time of the T(c)-division. In the third module, the expected development of ripple size across the T(c)-division is calculated using empirical relationships between rate of development of ripple height and grain-related mobility parameter, and subsequently compared with the observed development of ripple size in the turbidite. In this way, the accuracy of the accumulation time calculated in module 2 can be verified independently. TDURE was tested using Bouma-type turbidites from the Doheny Channel (Capistrano Fm., CA, USA) and the Flysch di Motta (Calabria, Italy). Accumulation times of 12 and 10.75 min for the T(c)-division, and 19 and 16 min for the T(abc)-sequence were calculated for the Doheny Channel and the Flysch di Motta turbidites, respectively. Although module 3 underestimates the rate of development of current ripples near the T(bc)-boundary in both beds, ripple size at the T(cd)-boundary is calculated accurately. The underestimation of development rate may be caused by differences between flow conditions in experimental flumes on which the model calculations are based and turbidity-current dynamics. (C) 2000 Elsevier Science B.V. All rights reserved.