A basic premise in marine heat flow studies is that the temperature gradient varies with depth as a function of the bulk thermal conductivity of the sediments. As sediments become more deeply buried, compaction reduces the porosity and causes an increase in the bulk thermal conductivity. Therefore, while the heat flow may remain constant with depth, the thermal gradient is not necessarily linear. However, it has been argued that measurements showing increased sediment thermal conductivity with burial depth may be caused by a horizontal measurement bias generated by increasing anisotropy in sediments during consolidation. This study reanalyses a synthesis of Ocean Drilling Program data from 186 boreholes, and investigates the occurrence of nonlinear geothermal gradients in marine sediments. The aim is to identify whether observed downhole changes in thermal conductivity influence the measured temperature gradient, and to investigate potential errors in the prediction of in-situ temperatures derived from the extrapolation of near-surface thermal gradients. The results indicate that the measured thermal conductivity does influence the geothermal gradient. Furthermore, comparisons between shallow measurements (<10 m) from surface heat flow surveys and the deeply constrained temperature data from 98 ODP boreholes indicate that the shallow gradients are consistently higher by on average 19 °C km–1. This is consistent with higher porosity and generally lower thermal conductivity in near-seafloor sediments, and highlights the need to develop robust porosity–thermal conductivity models to accurately predict temperatures at depth from shallow heat flow surveys.