Context. X-Ray bursts (XRBs) are powerful thermonuclear events on the surface of accreting neutron stars (NSs) where nucleosynthesis of intermediate-mass elements occurs. The high surface gravity of an NS prevents the ejection of material by the XRB thermonuclear explosion. However, the predicted and observed XRB luminosities sometimes exceed Eddington’s value, and some of the material may escape by means of stellar wind. Aims. This work aims to determine the mass-loss and chemical composition of the material ejected through radiation-driven winds and its significance for Galactic abundances. It also reports on the evolution of observational quantities during the wind phase, which can help constrain the mass-radius relation in NSs. Methods. A non-relativistic radiative wind model was implemented, with modern opacity tables and treatment of the critical point, and linked through a new technique to a series of XRB hydrodynamic simulations that include over 300 isotopes. This allowed us to construct a quasi-stationary time evolution of the wind during the XRB. Results. In the models we studied, the total mass ejected by the wind was about 6 × 1019 g; the average ejected mass per burst represented 2.6% of the accreted mass between bursts, with 0.1% of the envelope mass ejected per burst; and approximately 90% of the ejecta was composed by 60Ni, 64Zn, 68Ge, and 58Ni. The ejected material also contained a small fraction (10−4 − 10−5) of some light p-nuclei, but not enough to account for their Galactic abundances. Additionally, the observable magnitudes during the wind phase showed remarkable correlations, partly due to the fact that photospheric luminosity stays close to the Eddington limit. Some of these correlations involve wind parameters, such as energy and mass outflows, that are determined by the conditions at the base of the wind envelope. Conclusions. The simulations resulted in the first realistic quantification of mass-loss for each isotope synthesized in the XRB. The photospheric correlations we found could be used to link observable magnitudes to the physics of the innermost parts of the envelope, close to its interface with the NS crust. This is a promising result regarding the issue of NS radius determination.