This paper presents a statistical analysis of the circumgalactic environment of nearby Seyfert galaxies based on a computer-aided search of companion galaxies on the Digitized Sky Survey (DSS). We defined a sample of 72 nearby Seyfert 1 (redshift 0.007 ≤ z ≤ 0.034) and a sample of 60 Seyfert 2 galaxies (0.007 ≤ z≤ 0.020), which include only high galactic latitude objects. In addition, we built two control samples of non-active galaxies matching the number of sample members, the redshift, morphological type, and diameter distribution of the Seyfert 1 and Seyfert 2 samples separately. We stress how our sample selection introduces important methodological improvements that avoid several sources of strong bias. An intrinsic difference between the environment of Seyfert 1 and Seyfert 2 galaxies, suggested by previous work, is confirmed as statistically significant. For Seyfert 2 galaxies we find a significant excess of large companions (D C > ∼ 10 Kpc) within a search radius < ∼ 100 Kpc of projected linear distance, as well as within a search radius equal to three times the diameter D S of each Seyfert galaxy. For Seyfert 1 galaxies there is no clear evidence of any excess of companion galaxies neither within 100 Kpc, nor within 3D S . For all samples the number of companions actually counted within a search radius of 3 D S is a factor ≈ 2 above the expectation values derived from the number density of galaxies over one square degree fields centered on the sample galaxies, suggesting a markedly non-Poissonian distribution for galaxies on scales < ∼ 100 Kpc. This difference in environment is not compatible with the simplest formulation of the Unification Model (UM) for Seyferts: both types 1 and 2 should be intrinsicaly alike, the only difference being due to orientation of an obscuring torus. We propose an alternative formulation.