In recent years adherents of the neutral mutation hypothesis have conducted a variety of statistical tests concerning the applicability of their theory to data of biochemical genetic polymorphism in natural populations. From the other side, the involvement of natural selection as a main evolutionary force responsible for the observed levels of polymorphism and heterozygosity have also been proposed in numerous field studies and theoretical considerations. However, none of these hypotheses completely and satisfactorily explains the collected data. We believe that one of the main causes of the discrepancy in theories is that the variability at each locus is considered independently in both of the above-mentioned approaches. Yet, it was suggested long ago, and now there is an increasing amount of evidence indicating cooperation between different loci which can influence the variability at each of them. Thus we think that the use of models considering the genome as a suit of independent genes is a priori expected to decrease the efficacy of the approximation. In the present review we attempt to draw attention to findings of interdependence of the variability of different characters and its possible limiting action on the growth of genetic diversity in natural populations. We do not try to give a universal explanation for the processes acting in populations and determining levels of heterozygosity. However, to our mind, the solution of the discussed question requires consideration of genes in their interactions.