Abstract. Oxygen isotope (δ18O) sclerochronology of benthic marine molluscs provides a means of reconstructing the seasonal range in seafloor temperature, subject to use of an appropriate equation relating shell δ18O to temperature and water δ18O, a reasonably accurate estimation of water δ18O, and due consideration of growth-rate effects. Taking these factors into account, δ18O data from late Pliocene bivalves of the southern North Sea basin (Belgium and the Netherlands) indicate a seasonal seafloor range a little smaller than now in the area. Microgrowth-increment data from Aequipecten opercularis, together with the species composition of the bivalve assemblage and aspects of preservation, suggest a setting below the summer thermocline for all but the latest material investigated. This implies a higher summer temperature at the surface than on the seafloor and consequently a greater seasonal range. A reasonable (3 ∘C) estimate of the difference between maximum seafloor and surface temperature under circumstances of summer stratification points to seasonal surface ranges in excess of the present value (12.4 ∘C nearby). Using a model-derived estimate of water δ18O (0.0 ‰), summer surface temperature was initially in the cool temperate range (<20 ∘C) and then (during the Mid-Piacenzian Warm Period; MPWP) increased into the warm temperate range (>20 ∘C) before reverting to cool temperate values (in conjunction with shallowing and a loss of summer stratification). This pattern is in agreement with biotic-assemblage evidence. Winter temperature was firmly in the cool temperate range (<10 ∘C) throughout, contrary to previous interpretations. Averaging of summer and winter surface temperatures for the MPWP provides a figure for annual sea surface temperature that is 2–3 ∘C higher than the present value (10.9 ∘C nearby) and in close agreement with a figure obtained by averaging alkenone and TEX86 temperatures for the MPWP from the Netherlands. These proxies, however, respectively, underestimate summer temperature and overestimate winter temperature, giving an incomplete picture of seasonality. A higher annual temperature than now is consistent with the notion of global warmth in the MPWP, but a low winter temperature in the southern North Sea basin suggests regional reduction in oceanic heat supply, contrasting with other interpretations of North Atlantic oceanography during the interval. Carbonate clumped isotope (Δ47) and biomineral unit thermometry offer means of checking the δ18O-based temperatures.