Accurate quantification of total forest biomass requires sound estimates of root biomass. Because roots are challenging to measure in situ, data are sparse, and this has limited our understanding of root allocation across a range of forest types. Increased sampling in the oceanic and cool temperate forests of the southern hemisphere means we are now able to better understand biomass allocation in this biome. Here we compile and systematically review root:shoot ratios for these forests and examine the taxonomic, regional, and environmental determinants of variability in biomass allocation to roots. Specifically, we assess whether limited access to resources resulting from low temperatures or low rainfall:temperature ratios increases root allocation. A literature and database search identified 441 root:shoot records from 25 studies and 32 locations around the South Pacific. Records were weighted by excavated area as a measure of sampling effort. Root:shoot ratios for eucalypts in southeast Australia (0.277) and southern beech in South America (0.275) were 6-7% above the global mean of 0.26, while southern beech (0.233) and other dicots (0.234) in New Zealand were consistent with values reported for temperate (mostly northern hemisphere) broadleaved forests. Low root:shoot ratios (of trees >= 5 cm stem diameter) relative to the global mean were noted for South American gymnosperms (0.219), seemingly driven by low values in the Cupressaceae, and New Zealand tree ferns (0.194). Size effects were taxon dependent, with a marked decline in root:shoot ratios with increasing stem size for the southern beech in South America and modest effects for other combinations of taxa and region. A clear signal of increasing root allocation with declining rainfall:temperature ratios was detected, but no major effects of temperature were noted within the range of the data. Our findings support current approaches to biomass and carbon estimation but leave room for refinement, with specific recommendations given for the New Zealand context.