In the near future, new instruments such as ESPRESSO will arrive, allowing us to reach a precision in radial velocity measurements on the order of 10 cm s-1. At this level of precision, several noise sources that until now have been outweighed by photon noise will start to contribute significantly to the error budget. The telluric lines that are not neglected by the masks for the radial velocity computation, here called micro-telluric lines, are one such noise source. Aims. In this work we investigate the impact of micro-telluric lines in the radial velocities calculations. We also investigate how to correct the effect of these atmospheric lines on radial velocities. Methods. The work presented here follows two parallel lines. First, we calculated the impact of the micro-telluric lines by multiplying a synthetic solar-like stellar spectrum by synthetic atmospheric spectra and evaluated the effect created by the presence of the telluric lines. Then, we divided HARPS spectra by synthetic atmospheric spectra to correct for its presence on real data and calculated the radial velocity on the corrected spectra. When doing so, one considers two atmospheric models for the synthetic atmospheric spectra: the LBLRTM and TAPAS. Results. We find that the micro-telluric lines can induce an impact on the radial velocity calculation that can already be close to the current precision achieved with HARPS, and so its effect should not be neglected, especially for future instruments such as ESPRESSO. Moreover, we find that the micro-telluric lines' impact depends on factors, such as the radial velocity of the star, airmass, relative humidity, and the barycentric Earth radial velocity projected along the line of sight at the time of the observation.