Context : Masses are a fundamental parameter , but they are not well known for most hot subdwarfs .
In general , the mass of a hot subdwarf is derived with asteroseismology or dynamical methods , for which it is often difficult to obtain the necessary data from observations .

Aims : We intend to find an approach to deriving the masses of hot subdwarfs from observational data in the literature .

Methods : We presented full evolutionary calculations for hot subdwarfs in a wide mass range ( 0.33 M _ { \odot } to 1.4 M _ { \odot } ) for a Population I metallicity of Z =0.02 , and obtained a relation between M _ { p } and \log ( \frac { T _ { eff } ^ { 4 } } { g } ) , where M _ { p } , T _ { eff } , and g are the most probable mass , effective temperature , and gravity .
This relation is used to study the masses of some observed hot subdwarfs .

Results : We proposed a method of determining the masses of hot subdwarfs .
Using this method , we studied the masses of hot subdwarfs from the ESO supernova Ia progenitor survey and Hamburg quasar survey .
The study shows that most of subdwarf B stars have masses between 0.42 and 0.54 M _ { \odot } , whilst most sdO stars are in the range 0.40 \sim 0.55 M _ { \odot } .
Comparing our study to the theoretical mass distributions of Han et al .
( 2003 ) , we found that sdO stars with mass less than \sim 0.5 M _ { \odot } may evolve from sdB stars , whilst most high-mass ( > 0.5 M _ { \odot } ) sdO stars result from mergers directly .

Conclusions :