A duration of load study representing 13 years of testing was recently terminated. Preliminary results have been published over the years. This paper represents the final account of the study, which was focused on the influ-ence of moisture content on time to failure for structural timber subjected to bending under constant load condi-tions. Two constant moisture conditions (MC = 11 % and 20 %) and one condition of varying moisture (MC between 11 % and 20 %) were applied. A total of 816 Norway spruce boards of dimensions 44 x 95 x 1800 mm were included. Eight groups of non-destructively matched samples were formed. Four groups were subjected to short-term strength tests, and four groups were subjected to long-term tests. Creep and time to failure were moni-tored. Time to failure as a function of stress level was established and the reliability of stress level assessment was discussed. A significant mechanosorptive effect was demonstrated both in terms of increased creep and shortening of time to failure. The test results were employed for the calibration of four existing duration of load models. The effect of long-term loading was expressed as the stress level SL50 to cause failure after 50 years of loading. SL50 was found to be of the order 0.60 for MC = 11 %, 0.50 for MC = 20 % and 0.44 MC varying be-tween 11 % and 20 %. The test results revealed no evidence of a threshold stress level. A reliability based cali-bration of load-duration factors was performed using probabilistic models of loads and of the short-term and long-term strengths. For permanent and imposed library loads, reliability-based estimation of the load duration factor gave almost the same results as direct, deterministic calibration. Keywords: Creep, damage models, duration of load, equal rank assumption, load duration factor, matched sam-ples, probabilistic models, threshold level, time to failure.