Context : According to previous investigations , the effect of diffusion in the stellar atmospheres and envelopes of hot white dwarfs and subdwarf B ( sdB ) stars strongly depends on the presence of weak winds with mass-loss rates \dot { M } < 10 ^ { -11 } M _ { \odot } / yr .

Aims : As in most of these stars with luminosities L / L _ { \odot } \la 100 , no wind signatures have been detected , the mass-loss rates are unknown .
In the present paper mass-loss rates are predicted from the original theory of radiatively driven winds .

Methods : The method of solution is modified so that the usual parametrization of the line force multipliers is not necessary .
This is important especially for very thin winds .
In addition we checked whether a one-component description is justified .
As a consequence of various simplifications , the mass-loss rates are expected to be overestimated .

Results : Results are presented for effective temperatures in the range 25000 K \leq T _ { eff } \leq 50000 K and for various metallicities between solar and Z / Z _ { \odot } = 0.01 .
For ( pre- ) white dwarfs and sdB stars a stellar mass of M _ { * } = 0.5 M _ { \odot } is assumed .
For fixed values of T _ { eff } , M _ { * } , and Z , the results predict decreasing mass-loss rates with increasing surface gravity and an increasing dependence of the mass-loss rates on the metallicity .
For white dwarfs with \log g > 7.0 no wind solution exists even if the metallicity would be solar .
Winds with mass-loss rates around 10 ^ { -11 } to 10 ^ { -10 } M _ { \odot } / yr are predicted for the most luminous sdB stars with surface gravities of \log g \la 5.5 , if the metallicity is not significantly lower than solar .
For lower values of \dot { M } metals decouple from hydrogen and helium .

Conclusions : If weak winds with \dot { M } \la 10 ^ { -12 } M _ { \odot } / yr exist , the metals can not be coupled to hydrogen and helium .
This should lead to additional changes in the surface composition , which have not yet been taken into account in the diffusion calculations with and without mass-loss .
A possible scenario is the existence of pure metallic winds with mass-loss rates of \dot { M } \la 10 ^ { -16 } M _ { \odot } / yr and with hydrostatic hydrogen and helium .