Five data sets were used to estimate steric level fluctuations in the North Atlantic for 2003-2015. We compare estimates made by a combination of altimetry and GRACE gravity data (ALT-GRV) with assessments obtained from vertical density profiles derived from SODA reanalysis, ARMOR, and EN4 objective analyses. We analyze the datasets without linear trends, and the seasonal signals are also removed. The resulting signals demonstrate the steric sea-level anomalies not related to the linear trends and the seasonal cycles and can be connected with short-period intra-annual variability as well as vortex dynamics of the region since mesoscale eddies can transfer heat and salt and influence thereby the thermohaline water structure from the sea surface to the depth. The deep convection, as well as meandering of the currents also influences the variability of residual time series. The steric sea-level fluctuations, obtained from the ARMOR dataset, which incorporates results of satellite observations, shows the best fit for those, derived from ALT-GRV data. The correlation coefficient between ARMOR and ALT-GRV varies between 0.6 and 0.8 over the study region (0.7 on average). Steric sea-level variations derived from SODA or EN4 show good matches with ALT-GRV only for the steric sea-level fluctuations spatially averaged over central regions of the North Atlantic. The discrepancies between the data sets increase northwards and towards the coast. Of the considered data sets, ARMOR is the most suitable for climate studies and research of the sea-level change effects; however, it should be used with caution in the study of the spatial distribution of the steric level.