Random frequency drift causes linkage disequilibrium and associative overdominance, and they have important bearing on the investigation of fitness values of marker alleles. Very often, experimental results merely reflect the effects of genetic background rather than the effect of the markers per se. Approximate magnitude of linkage disequilibrium and associative overdominance were estimated assuming various situations in natural populations. They mainly depend on N_ec, the product of the effective population number and the recombination fraction. Also, the magnitude of associative overdominance was estimated assuming two models of experimental populations. In one model, a neutral marker gene is linked with a number of mildly deleterious mutants, and in the other it is linked with overdominant loci. Based on the results, the importance of associative overdominance in both experimental and natural populations was discussed.