A model for time-to-failure prediction based on component parameter drift is described. The idea for creation of this model is based on the influence of time-dependent random and non random factors on the distribution of the random variable.The reliability interpretation of the data from thick film resistor ageing tests has been completed with the model developed. Two different drift functions are described for two ruthenium-based thick film methods with equal resistivity – 10Ω/square. The values of the functional parameters depend strongly on the storage test temperature. Mean time-to-failure decreases approximately 4 times with an increase of storage temperature of 20 deg. C.