ABSTRACT:A modification in the rate of change of sea level (i.e. an 'acceleration' or 'nonlinear trend') is an important climate-related signal, which requires confirmation and explanation. In this study, the evidence for accelerations in regional and global average sea level on timescales of several decades and longer is reviewed by inter-comparison of the recent findings of different researchers and by inspection of original tide gauge records. Most sea-level data originate from Europe and North America, and both the sets display evidence for a positive acceleration, or 'inflexion', around 1920-1930 and a negative one around 1960. These inflexions are the main contributors to reported accelerations since the late 19th century, and to decelerations during the mid-to late 20th century. However, these characteristic features are not always found in records from other parts of the world. Although some aspects of the sea-level time series are consistent with changes in rates of globally averaged temperature changes, volcanic eruptions and natural climate variability, modelling undertaken so far has been unable to describe these features adequately. This emphasizes the need for a major enhancement of the sea-level data set, especially for those parts of the world without long tide gauge records, in order to obtain greater insight into the spatial dependence of accelerations. A number of complementary methods must be employed, of which salt marsh techniques offer the possibility of obtaining time series similar to those that would have been obtained from coastal tide gauges.