Abstract Estimates of extreme sea levels and return periods have been based mainly on hourly sampling rates. Technological development has enabled the sampling rates to increase and sampling rates of 5–10 min are becoming increasingly common. In this paper we explore the relationship between extreme sea levels and estimated return periods based on hourly and shorter sampling periods in three tide-gauges one at the Atlantic coasts of Spain (Coruña), one in the western Mediterranean (Malaga) and one in the N. Adriatic (Trieste). Significant differences of several centimetres are found in the hourly and 5 min extremes. These reflect in significant underestimation of the 50-year return levels which in Trieste reach 38 cm. A theoretical relationship between the high and the low sampling rate of extremes is also tested. Thus updated 50-year return levels for the Mediterranean and the coasts of the Iberian peninsula are produced assuming that the differences identified in the various stations generalise to other tide-gauge (hourly) records for which hourly values have been analysed earlier.