Abstract Pleistocene cyanobacterial mounds composed principally of fresh water branched cyanobacterium Rivularia haematites at the west end of Perachora Peninsula, eastern Gulf of Corinth, grew in water on an eroded terrace in a location where several fault systems converge near the triangular end of the peninsula, surrounded by deep water. Differential uplift due to faulting cut many mounds, which were eroded leading to caves inside mounds. Cave-fill stratigraphy reveals subsequent marine submergence, then emergence, then final marine submergence which buried mounds in MIS (Marine Isotope Stage) Stage 5e marine corals and shells. Coralline algae comprising only the euryhaline species Lithophyllum pustulatum postdate the mound frame (between branches in the mounds, and as a previously undescribed pendent growth form with microbialites as cryptic biota in caves). The mound growth phase presents a conflict of interpretation between lacustrine and marine origins. Lacustrine interpretation is favoured by mound ecology (domination by fresh water cyanobacterium Rivularia haematites ) when the Gulf of Corinth was a lake (MIS Stage 6 lowstand); and euryhaline coralline alga Lithophyllum pustulatum grew only after the R. haematites growth phase was complete, presumably when marine influence developed in early 5e. However, published evidence of geochemical signatures suggest the mounds grew in the MIS 5e sea during a phase of intense expulsion of carbonate- and calcium-rich groundwaters from local faults, which diluted the sea water sufficiently to permit calcification of R. haematites in a marine setting. That scenario requires coincidence of: location of the terrace in shallow water during 5e; calcification of a fresh water cyanobacterium during only the first 5e sea-level peak, ceased before the late 5e peak and was never repeated. Lacustrine interpretation requires non-marine calcification, but also needs a subsequent relative lake-level fall to form caves before 5e marine flooding; sea-level fluctuations within 5e could be responsible for formation of cave fill before final burial. Both interpretations are influenced by uplift rates, with the published value of c. 0.3 mm a -1 favouring the marine interpretation in Stage 5e. However, a lacustrine mound origin can be accommodated by a phase of rapid uplift in late Stage 6 to early Stage 5, then fluctuating sea-level in Stage 5e during slow uplift in a complex fault setting; after late Stage 6 to early Stage 5e, uplift rate slowed and the mounds were probably continuously exposed after 5e. For the lacustrine interpretation the debate continues about whether or not: (a) the marine-affinity geochemical signatures in R. haematites are due to alteration by 5e sea water; and (b) a variable uplift rate applies in this complex fault conjunction. However, the marine interpretation seems to require a powerful and sustained expulsion phase of groundwater across the entire end of Perachora Peninsula to dilute sea water sufficiently over a large area throughout the mound growth phase (possibly as much as 2000 years) to allow formation of the Rivularia mounds and exclude all marine biota, a phase that was stopped abruptly and not repeated. This is considered more difficult to accommodate and, therefore, pending further evidence, the non-marine interpretation is favoured.