In this table we report the observing log for the GJ3998 spectra and the radial velocities, S, and Hα indexes. The star GJ3998 has been monitored from BJD=2456439.6 (26 May 2013) to BJD=2457307.8 (12 October 2015). We obtained a total of 136 data points spanning 869-days. The spectra were obtained at high resolution (R=115000) with the optical echelle spectrograph HARPS-N with exposure times of 15 minutes and an average signal-to-noise ratio (S/N) of 45 at 5500Å. Of the 136 epochs, 76 were obtained within the GAPS time and 60 within the Spanish time. Observations were gathered without the simultaneous Th-Ar calibration, which is commonly used to correct for instrumental drifts during the night. The M-type stars of the HADES program were observed by the Italian team in conjunction with other GAPS targets, which used the Th-Ar simultaneous calibration, therefore we estimated the drift data between the two fibers (star and reference calibration) for each night from these observations and evaluated the interpolated drift for GJ3998 (0.7m/s). <P />Data reduction and spectral extraction were performed using the Data Reduction Software (DRS, Lovis & Pepe, 2007A&A...468.1115L, Cat. J/A+A/468/1115). RVs were measured by means of a weighted cross-correlation function (CCF) with the M2 binary mask provided with the DRS. The RVs were also measured by matching the spectra with a high S/N template obtained by coadding the spectra of the target, as implemented in the TERRA pipeline (Anglada-Escude & Butler, 2012ApJS..200...15A, Cat. J/ApJS/200/15), which provides a better RV accuracy when applied to M dwarfs. We list the observation dates (barycentric Julian date or BJD), the signal-to-noise ratios (S/Ns), the radial velocities (RVs) from the DRS and TERRA pipelines (indicated with a T) and the Hα and S indexes, calculated both by the TERRA pipeline and by an independent method described in the text. The RV errors reported are the formal ones and do not include the jitter term. The S index and Hα errors are calculated as described in the text and do not take into account the photon noise. The S index and Hα errors derived from the TERRA pipeline are due to photon noise through error propagation. (1 data file).