The magnetic topology, structure, and geometry of the magnetic obstacles embedded within interplanetary coronal mass ejections (ICMEs) are not yet fully and consistently described by in situ models and reconstruction techniques. The main goal of this work is to better understand the status of the internal magnetic field of ICMEs and to explore in situ signatures to identify clues to develop a more accurate and reliable in situ analytical models. We take advantage of more than 20 years of Wind observations of transients at 1 AU to compile a comprehensive database of ICMEs through three solar cycles, from 1995 to 2015. The catalog is publicly available at and is fully described in this article. We identify and collect the properties of 337 ICMEs, of which 298 show organized magnetic field signatures. To allow for departures from idealized magnetic configurations, we introduce the term “magnetic obstacle” (MO) to signify the possibility of more complex configurations. To quantify the asymmetry of the magnetic field strength profile within these events, we introduce the distortion parameter (DiP) and calculate the expansion velocity within the magnetic obstacle. Circular-cylindrical geometry is assumed when the magnetic field strength displays a symmetric profile. We perform a statistical study of these two parameters and find that only 35% of the events show symmetric magnetic profiles and a low enough expansion velocity to be compatible with the assumption of an idealized cylindrical static flux rope, and that 41% of the events do not show the expected relationship between expansion and magnetic field compression in the front, with the maximum magnetic field closer to the first encounter of the spacecraft with the magnetic obstacle; 18% show contractions (i.e. apparent negative expansion velocity), and 30% show magnetic field compression in the back. We derive an empirical relation between DiP and expansion velocity that is the first step toward improving reconstructions with possible applications to space weather studies. In summary, our main results demonstrate that the assumed correlation between expanding structure and asymmetric magnetic field is not always valid. Although 59% of the cases could be described by circular-cylindrical geometry, with or without expansion, the remaining cases show significant in situ signatures of departures from circular-cylindrical geometry. These results will aid in the development of more accurate in situ models to reconcile image.