We have shown that counting the domain-to-protein links observed in the protein and protein domain/motifs data sets and analysis of statistical distributions of these counts lead to simple probabilistic model of evolution of proteome complexity of the archael, bacterial and eukaryotic organisms [1-3]. In this work using InterPro data sets, we test the basic assumptions of our model. We conclude that the domain occurrence counts in a proteome during evolution is a random birth-death quasi-steady state process such that new domains are rarely appeared as singletons and lost at constant rates, and domains are reused and lost at rates proportional to their current use.