Volcanic soils can effectively store organic carbon (OC) for time scales of 10^2 to 10^4 of years. Previous work suggests that carbon-mineral interactions play a key role in the long term persistence of OC. This dissertation examines a transect on Kohala Mountain (Island of Hawaii), where soils have formed on a 400 ka basaltic substrate. Along the transect, trade winds drive a precipitation gradient leading to systematic changes in soil moisture, pH, and mineralogy. To examine the persistence of the SOC as soil conditions change, I utilized a series of inorganic and organic geochemical techniques, including metal analysis, carbon analysis, and pyrolysis-oxidation experiments. I used high resolution radiocarbon measurements on bulk soils, thermal fractions collected from ramped pyrolysis/oxidation (RPO), and compound-specific lipid biomarkers to measure the complex differences in the age from bulk to compound scale. This approach allowed for the interpretation of incremental changes in soil carbon energetics and age along the climate gradient. Soil organic horizons show modern radiocarbon values across the gradient, while deeper mineral soils yield variable but generally significantly older radiocarbon ages. In the wetter part of the climate gradient the soils are significantly depleted in iron oxides. The Fe-poor mineral subsoils have much younger 14C ages than in equivalent soils that retain Fe. Activation energy distributions p(0,E) from RPO data are calculated in each sample. I found that while the age gradient of the thermal fractions was relatively flat in all samples, the activation energies between organic horizons and mineral horizons differed. Mineral interactions dominate the activation energy signal resulting in a shift to a lower activation energies in subsoils verse organic horizons. This indicates two distinct mechanisms of contributing to the OC persistence across the climate gradient. The increase in activation energy in the organic horizons indicates an increased temperature sensitivity of this younger material, while mineral surface interaction decreases the OCs vulnerability to degradation. Lipid biomarkers were used to constrain the end members of SOC at these sites. Long-chain n-alkanoic acids were older than both RPO fractions and bulk 14C values. This effect was site-specific, so irrespective of actual age, plant waxes are much younger in soils where Fe is no longer abundant indicating Fe plays an important function in OC stabilization in volcanic soils across scales, from single compounds to bulk OC. Over all I find that mineral weathering has a strong effect on the persistence of soil OC. The presence of Fe minerals in subsurface mineral horizons is associated with carbon that is stabilized for > 10,000 years. The data are consistent with a major role for mineral stabilization of old organic carbon in these volcanic soils. The role of oxidized iron minerals appears to be particularly important. Because the stability of Fe-oxides is climate dependent, the storage of large amounts of old carbon in deep soil mineral horizons is potentially vulnerable to climate change.