The 10 m-thick Holocene succession at core site M02-45 (41 degrees 41.17'N, 28 degrees 19.08'E, -69 m water depth) has key advantages for investigation of the Sr-isotopic evolution of the Holocene Black Sea. Earlier studies have focussed on thinner successions and coquinas. At the M02-45 site (augmented by nearby core M05-03P),Sr-87/Sr-86 determinations on mollusc shells extracted at mostly 10-20 cm depth increments provide a temporal resolution mostly < 200 years in sediment older than similar to 5500 cal yr BP, and similar to 20-25 years for the early Holocene reconnection between the Black Sea (formerly the Neoeuxine Lake) and the global ocean. Isotopic measurements are in stratigraphic order, so temporal trends are unambiguous. Measurements were made using a Neptune multi-collector inductively coupled plasma mass spectrometer (MC-ICPMS) and have 1 sigma uncertainties of similar to +/- 0.000015. There are four stages of Sr-87/(86) Sr increase and salination associated with the reconnection. From 12,145-9580 cal yr BP (stage A), before first Mediterranean inflow, the Sr-isotopic ratio varied from 0.708847-0.708881. Modelling suggests that the Sr concentration in the Neoeuxine Lake might have been several times higher than modern river values because of evaporative concentration. For similar to 100 years immediately before reconnection (9580-9490 cal yr BP, stage B),Sr-87/(86) values dropped to their lowest levels: 0.708841-0.708843. Abruptly (in geological terms),Sr-87/(86) then began to climb starting 9465-9490 cal yr BP and reached a quasi-steady-state 'plateau' with ratios similar to 0.708965 by similar to 9380 cal yr BP (stage C). The sharp Sr-87/(86) increase marks the first significant intrusion of saline water into a previously isolated Neoeuxine Lake. The quasi-steady-state condition lasted 350-400 years. Subsequently, starting similar to 8985 cal yr BP and proceeding to the present day, there was a step-wise rise of Sr-87/(86) to modern levels (stage D), during which a salinity threshold was passed that allowed widespread replacement of brackish-water faunas by Mediterranean species. Modelling suggests that the lake/sea level likely did not, and could not, rise from -120 m to - 30 m between 9490 and 9380 cal yr BP unless (a) the Sr concentration in the pre-reconnection Neoeuxine Lake was 3-4 times higher than today, or (b) the water column was strongly stratified during first entry of saline water. The second alternative is very unlikely because of seasonal vertical mixing (downwelling/upwelling) in what was then a rather homogeneous temperate lake. Catastrophic flooding of a lowstand lake would require an average discharge through the Strait of Bosphorus of similar to 9500 m(3)s(-1), whereas saline entry of Mediterranean water as an underflow into an already high lake could reproduce the first stage of Sr-87/(86) increase with an average discharge as low as similar to 2200 m(3)s(-1). Because the M02-45 site is similar to 50 m above the late Pleistocene lowstand shoreline and contains sub-wavebase sediments with Sr-87/(86) values that record the first entry of saline water into the Neoeuxine Lake, the surface of the lake must have been significantly higher than - 70 m at the time of the reconnection. Two prominent 'plateaux' which punctuate the long-term Sr-87/(86) increase are attributed to decadal to centennial periods of increased discharge from European rivers, creating a positive hydrological balance and effectively blocking or seriously impeding saline-water advance up the Strait of Bosphorus toward the Black Sea.