We reassess whether range shifts in marine organisms are occurring in a direction predicted by causation due to climate using two different methods: a global meta-analysis and a regional analysis of museum records focussing on marine molluscs. Globally we used rigorous selection criteria to investigate studies describing range shifts in marine organisms, including exclusion of single-species studies and those that inferred range shifts through shifting abundance. For each species meeting these criteria (311 species from 13 studies), the distance in kilometres moved per year was calculated, and life history characteristics such as mobility, habitat, dispersal potential and trophic level were recorded to assess if these were linked to the shifts. We also examined the potential correlation between changing sea surface temperatures and range shifts. Regionally we analysed the intertidal mollusc collection of the Australian Museum and Museum of Victoria and consulted with mollusc experts to determine if museum records could be used to detect range shifts in marine invertebrates. The global meta-analysis showed that individual species shifted on average 8.9 km polewards per year. Range shifts were not significantly related to temperature change or life history characteristics with the exception of intertidal and nearshore subtidal fauna showing poleward shifts (6.8 and 14.3 km per decade, respectively) and shelf fauna showing a slight equatorial shift (2.7 km per decade). Our regional analysis showed that 54 of the 634 mollusc species examined shifted their range into or out of the well-surveyed Sydney region. However, closer examination by mollusc experts revealed that the evidence for these shifts was unreliable, and we conclude that caution is required when using museum records in range shift studies. Overall, our study supports previous research showing a poleward shift in response to climate change, although this is an order of magnitude less than values reported in the only other meta-analysis on marine range shifts, possibly owing to the strict selection criteria applied here.