The main sequence mass functions (MF) of a large sample of Galactic clusters (young and old) can be well reproduced with a tapered power law distribution function with an exponential truncation of the form dN/dM ∝ m-α[1 - exp(-m/mp)β]. The average value of the power-law index α is very close to Salpeter (~2.3), whereas the peak mass mp is in the range 0.1-0.6 M⊙ and does not seem to vary in a systematic way with the present cluster parameters such as metal abundance and central concentration. A remarkable correlation with age, however, is seen in that older disc clusters have higher mp, although this trend does not extend to globular clusters, whose value of mp is lower than that of old open clusters. This trend most likely results from the onset of mass segregation following early dynamical interactions in the loose cluster cores. Differences between globular and younger clusters may depend on the initial environment of star formation, which in turn affects their total mass. Mass functions of field populations such as the solar neighbourhood and bulge are consistent with the hypothesis that they were built up over time by contributions from many functions of this type with different peak masses.