The latitudinal diversity gradient (LDG) is a phenomenon acknowledged for over two centuries. The LDG of marine crustaceans has been studied often but without reaching consensus on its ultimate causative processes. We have undertaken a new synthesis to assess the generality of the LDG and evaluated how potential sampling and other biases, spatial scale, geographic regions, taxonomic aggregation, and differences between clades affect patterns. A meta-analysis of 186 datasets, encompassing 20 studies and 7 crustacean orders, revealed a strong effect size of the species richness-latitude correlation, supporting the existence of a “canonical” LDG. The effect size was sensitive to spatial scale, with studies conducted over shorter latitudinal ranges tending to show a weaker LDG. Correcting for sampling biases in the number of occurrences, taxonomic completeness and spatial heterogeneity did not affect the strength of the LDG, nor did the degree of taxonomic aggregation; effect sizes were similar at family and ordinal levels. However, between orders effect sizes varied strongly, with peracarid orders (Amphipoda, Cumacea, Isopoda) showing a weaker or inverse LDG compared with non-peracarid orders (Calanoida, Euphausiacea, Decapoda, Sessilia). Additional analyses based on a global dataset of >2 million occurrences of >13,000 species revealed patterns undetected by the meta-analysis, including: (1) the existence of a marked bi-modal LDG, with peaks of diversity in subtropical areas (Calanoidea, Decapoda, Sessilia) and in temperate areas (Amphipoda, Isopoda), (2) interhemispheric asymmetry, variable across groups and depths, and (3) ocean basin differences in the shape of the LDG, dependent on taxonomic clade. Both ecological and evolutionary processes play a part. The fossil record of Decapoda showed that its global canonical LDG can be explained by median and range of the age of genera, i.e., hotspots of diversity harbor both younger and older genera and contain a high proportion of genera originating during the Paleogene. In addition, the effect size was negatively related to family age, the LDG being stronger in older families of early Cenozoic and Mesozoic origin. Modes of larval development also played a significant part, taxa without planktonic larvae having weaker or inverse LDG compared with taxa with pelagic larvae. Because clades with direct development tend to show smaller bathymetric and latitudinal ranges than those with pelagic larvae, differences in diversification rates may be implied. Overall, our evidence suggested that the ultimate causes of the LDG are deeply tied to geographic differences in macro-evolutionary rates, i.e., greater rates of species origin and lower rates of extinction in the tropics than in higher latitudes combined with a strong tropical niche conservatism.