Many steel rope systems are subjected to fluctuating tensile loads and therefore can fail due to fatigue. Available fatigue test data indicate that rope diameter, mean stress, socket type, lay angle and rope length influence the fatigue resistance. Most of the tests were terminated before full failure of the ropes. This paper shows that the test termination criterion, such as fracture of the first wire, fracture of 5 % of the wires or full rope fracture, has a large influence on the resulting fatigue resistance. A probabilistic analysis is carried out for a rope system in a bridge, demonstrating that the required structural reliability levels are met when considering full failure as the end-of-life criterion for ropes.