A large body of experimental data consisting of 116 samples (sets) of individual life span (LS) values of D. melanogaster from the same laboratory strain Canton-S was analyzed. In total, 10180 Drosophila flies (5100 females and 5080 males) were studied. Each of 58 pairs of samples belonged to a definite generation in a continuous succession where every next generation was an offspring of the preceding one. Mathematical simulation made it possible to demonstrate that both the normal (Gaussian) and Gompertz distributions were equally good approximations of the experimental data. Both of them adequately described the LS distributions in laboratory populations of D. melanogaster. The confidence intervals for absolute deviations of the theoretical distributions from experimental ones were small (4-5%). In other words, the approximation error was no more than 5% in either case. The estimation of the dependence of approximation quality on the LS in the original (experimental) population showed that the normal distribution was preferable, because, in this case the absolute deviation from the experimental distribution was independent of the LS in the original population.