The article refers to real-time systems, especially being used in industry as IT systems with time constraints. A typical analysis of such systems is based generally on the maximum values, which can be useful to determine the time characteristic in worst cases of system run. The proposed method is based on all values being available to measure used in the form of probability mass functions. This method uses the convolution and deconvolution of temporal characteristics of system sub-processes. It is shown that this approach can be used to verify the assumption that the maximum response time of a given complex system is the sum of maximum response times of its parallel executed sub-processes.