An extensive database was created involving sea surface temperature reconstructions from the Atlantic Ocean of the Holocene period. The records from five different regular applied proxy methods have been included. The five proxies are alkenone unsaturation, Mg/Ca paired with ä18O (planktonic foraminifera), and faunal assemblages of dinocysts, diatoms and planktonic foraminifera. The first two methods are used to reconstruct annual temperatures, while the faunal methods are used for summer and/or winter temperature reconstructions. There are four parts discussed in this thesis: (1) a comparison between the five proxies; (2) the temperature Holocene and the related oceanic circulation changes; (3) centennial timescale analysis of the data in relation to solar activity; (4) the comparison of proxy data with the ECBILT-CLIO-VECODE model. A study of the differences in average Holocene temperatures, temperature evolution, and temperature variability between the proxy records reveals that there are some definite trends. Seasonal preference in productivity and growth season is considered to be the best explanation for anomalously high alkenone temperatures records toward higher latitudes. Deviating temperature trends and a lack of seasonal variability in planktonic foraminifera based records (faunal and Mg/Ca) from particularly mid and high latitudes suggest the influence of thermocline temperatures. This is consistent with the analysis that the average dwelling depth for planktonic foraminifera is deeper than for the other proxies. These findings thus suggest that particularly at high latitudes, the interpretation of at least three of the five proxy methods should be reconsidered. Four distinct Holocene periods could be extracted from the records. A first period between roughly 11,5 and 9,5 ka BP is associated with a lot of temperature variability in the higher latitudes and a warming trend in practically the whole Atlantic region. The temperature increase and high variability have been associated with the start of meridional overturning circulation in the area between Greenland Iceland and Norway area and instability in the thermohaline circulation. The second period occurs between about 9,5 and 8 ka BP. This period is characterized by a small decrease in temperatures in most areas of the North Atlantic region (0,5-1,5°C), while the NW Atlantic experiences a sudden temperature rise. This temperature rise is most distinct (about 6°C) in the area west of the Reykjanes Ridge. A possible cause for this sudden replacement in heat distribution is a northward shift of the Gulf Stream due to high summer insolation and a high temperature gradient between eastern Canada and the warm Atlantic waters. Around 8 ka BP there is again a sudden change in temperature evolution. The 8,2 ka BP event and the initiation of Labrador Sea Water formation are most likely the triggers for this temperature shift (de Vernal et al., 2006). The period between about 8 and 5,5 ka BP is marked by an opposing temperature trend in the North Atlantic compared to the second period. The initiation of Labrador convection and a decrease in summer insolation probably lead to a southward shift of the Gulf Stream (Sachs, 2007). Decreasing summer insolation and the melt 2 | B e r n d R o m b a u t of the last part of the Laurentide Ice Sheet probably lead to the last major reorganization of the oceanic and atmospheric circulation around 5,5 ka BP. The major temperature shift at about 5,5 ka BP could be related to a threshold in the southward shift of the Gulf Stream. The last period between 5,5 ka BP until preindustrial times, is a period of general temperature decrease in practically the whole Atlantic region. The most pronounced temperature decrease during the last 8 kyr is in the NE Atlantic. This and some other changes suggest a trend towards a positive NAO. Analysis of the centennial variability of the records led to the discovery of periodicities of 150-200 year and about 500 year. Both periodicities indicate a relation with solar activity. The 150-200 year cycle probably corresponds to the ~205 year de Vries cycle. A comparison of the 500 year cycle with solar activity minima also suggests a strong similarity between the two. The ECBilt-CLIO-VECODE model is able to estimate average temperature changes over a broad latitudinal band, but it fails picking up local changes. The fact that the model does not simulate the southward shift in the Gulf Stream, probably leads to the anomalously high temperatures in the Atlantic region between particularly 50 and 60°N. From all the models in the PMIP2 database, only the CSIRO-Mk3L-1.1 model seems to simulate the southward shift in the Gulf Stream.