A sampled-continuous probability method of velocity measurement of an object having informatively-structural inhomogeneity by periodical execution of a cycle of the measurement, with determination of time during which a selected local part of the object moves along the measuring base of zone of observation is proposed. Selection of the local part in the beginning of the measuring base and its subsequent determination at the end of the measuring base is performed with the use of an informative aggregate of selected informative signs of a dominant informative component of the inhomogeneity. In order to increase the accuracy and speed of the method, the beginning of each cycle of velocity measurement is performed in response to a special command formed after the beginning of a preceding cycle of the velocity measurement with a given interval of time of shifting of the cycles, but within the preceding cycle. In each cycle of the velocity measurement a total time of execution of last stages connected with the computation and saving of the measuring information is selected smaller than minimum possible working step of time quantization of the velocity to be measured. Probability estimates of various time characteristics of the dominant informative component are used to provide invariability of metrological characteristics of the method to changes of the velocity to be measured within a wide dynamic and frequency range.