This table contains the second Swift X-ray Point Source (2SXPS) catalog of detections by the Swift X-ray Telescope (XRT) used in Photon Counting (PC) mode in the 0.3-10 keV energy range. Swift is a NASA mission with international participation dedicated to the gamma-ray burst study. It carries three instruments. The BAT is the large field-of-view instrument and operates in the 10-300 keV energy band; and two narrow field instruments, XRT and UVOT, that operate in the X-ray and UV/optical regime, respectively. The overall 2SXPS catalog characteristics are as follows: <pre> Data included 2005 Jan 01 - 2018 August 01 Sky coverage 3,790 square degrees Typical Sensitivity (0.3-10 keV) 2x10^-13 erg cm^-2 s^-1 (observations) 4x10^-14 erg cm^-2 s^-1 (stacked images) Typical position error 5.6" (90% confidence radius, including systematics) Detections 1,091,058 Unique sources 206,335 Variable sources 82,324 Uncatalogued sources 78,100 False positive rate Flag=Good 0.3% Flag=Good/Reasonable 1% Flag=Good/Reasonable/Poor <10% </pre> This catalog enhances the 1SXPS catalogue (Evans, P. A., et al. 2014, ApJS, 210, 8) in different ways. The 2SXPS catalog uses an improved Point Spread Function (PSF) and pile-up models, a better source detection pipeline that includes a technique to model the effects of stray light, and tests to automatically avoid diffuse emission and ~six years more data. The results are that the 2SXPS catalog contains 50% more temporal coverage than 1SXPS, a sky coverage of 3790 square deg almost double compare to the 1SXPS (1905 square Degree) and ~30% more sources compared to the 1SXPS. The Swift XRT observations were filtered to remove times when: a) data were contaminated by scattered light from the daylight side of the Earth; b) the on-board astrometry derived from the images obtained by the Swift UV/Optical telescope was unreliable; and c) observations with less than 100s of PC mode. The 127519 observations included in the catalog provide a total usable exposure is 266.5 Ms. A Swift observation is a collection of snapshots and the source detection algorithm was run on individual observation as well as on stacked images. The latter were generated on a grid of 2,300x2,300 pixels (~ 90'x90') to ensure that every overlap between observations is in at least one stacked image. A total of 14628 stacked images were generated. Each record corresponds to a unique source which characteristics are described with 230 parameters. The catalog reports for each source rates in four energy band (0.3-10.keV, 0.3-1. keV, 1-2 keV and 3-10 keV), background rates, variability for each energy band, two hardness ratio, peak rate and several spectral parameters. The hardness ratios are defined as follows: <pre> HR1 = (M-S)/(M+S) where M and S are the medium (1-2 keV) and soft (0.3-1 keV) band count rates HR2 = (H-M)/(H+M) where H and M are the hard (2-10 keV) and medium (1-2 keV) band count rates </pre> and they are calculated using all observations. The peak rate is determined using three different timescale: the count rate considering all the observations (see parameters rates in this database), the count rate in each observation (not reported in this database) and the count rate in each snapshot (not reported in this database). The peak rate is the rate +/- error from the timescale which has the highest 1-sigma lower-limit on the count rate. Spectral parameters and source flux are estimated using three different methods for two spectral models, a power-law and APEC (see Smith et al., 2001, ApJL, 556, L91). Not all sources have values for all three methods. The parameters starting with "fix" are defined for every source and uses fixed spectral model parameters: a photon index of 1.7 for a power-law model, a temperature of kT=1keV for the APEC model and for both models uses the Galactic absorption listed in the parameter "nh". The parameters starting "intr" have been inferred from