ABSTRACT The first and second Gravitational Wave Transient Catalogs by the LIGO/Virgo Collaboration include 50 confirmed merger events from the first, second, and first half of the third observational runs. We compute the distribution of recoil kicks imparted to the merger remnants and estimate their retention probability within various astrophysical environments as a function of the maximum progenitor spin (χmax), assuming that the LIGO/Virgo binary black hole (BBH) mergers were catalyzed by dynamical assembly in a dense star cluster. We find that the distributions of average recoil kicks are peaked at about $150\, \rm km\, s^{-1}$, $250\, \rm km\, s^{-1}$, $350\, \rm km\, s^{-1}$, $600\, \rm km\, s^{-1}$, for maximum progenitor spins of 0.1, 0.3, 0.5, 0.8, respectively. Only environments with escape speed ${\gtrsim}100\, \rm km\, s^{-1}$, as found in galactic nuclear star clusters as well as in the most massive globular clusters and super star clusters, could efficiently retain the merger remnants of the LIGO/Virgo BBH population even for low progenitor spins (χmax = 0.1). In the case of high progenitor spins (χmax ≳ 0.5), only the most massive nuclear star clusters can retain the merger products. We also show that the estimated values of the effective spin and of the remnant spin of GW170729, GW190412, GW190519_153544, and GW190620_030421 can be reproduced if their progenitors were moderately spinning (χmax ≳ 0.3), while for GW190517_055101 if the progenitors were rapidly spinning (χmax ≳ 0.8). Alternatively, some of these events could be explained if at least one of the progenitors is already a second-generation BH, originated from a previous merger.