A large-scale seafloor observation network, known as S-net, has been operated by National Research Institute for Earth Science and Disaster Resilience (NIED) in the Japan Trench area since 2016. The main objective of the observation is to provide reliable earthquake and tsunami early warnings. The network consists of 150 in-line-type observatories with sensors housed in cylindrical pressure vessels. The stations and cables were buried about one meter beneath the seafloor in the region of water depth less than 1500 m while they were sited on the seafloor in the deeper area. Previous studies have shown that the S-net sensors rotate after some threshold peak ground accelerations due to poor coupling between the sensor house and sediment, resulting in large rotational noises. Another serious concern at the S-net sites is the amplification effect of sediments. To minimize the various effects, researchers have devised equations using the displacement amplitude of vertical component of motions to estimate the magnitude for Japan Meteorological Agency earthquake early warning (Hayashimoto et al. 2019). The vertical motions are affected as well by the presence of water layer in the oceanic area. Thus, the analysis of seafloor records is challenging due to the multiply effects of sediments and water layer. In this meeting, we report the results of analyses of strong-motion records at the ground in the Kanto Basin and adjoining areas and the S-net sites from nearby moderate to strong offshore earthquakes. Strong motions at ground and borehole sites were taken from K-NET/KiK-net and MeSO-net, respectively. We have found that the horizontal PGAs and PGVs at the S-net sites were, on average, similar to or larger than those at the ground and borehole sites at equal distances. In contrast, the vertical PGAs and PGVs at the S-net sites tended to be smaller than those at the ground sites for the S wave. Notably, the PGAs and PGVs for the P-wave parts on the vertical records of S-net were smaller than those at the ground sites. A portion of the results was published in Dhakal and Kunugi (2021). We are now analyzing a larger dataset and report the detailed results elsewhere. This study was supported by the Advanced Earthquake and Tsunami Forecasting Technologies Project of NIED and JSPS KAKENHI Grant Number JP20K05055.