Many important questions in ecosystem science require estimates of stocks of soil C and nutrients. Quantitative soil pits provide direct measurements of total soil mass and elemental content in depth-based samples representative of large volumes, bypassing potential errors associated with independently measuring soil bulk density, rock volume, and elemental concentrations. The method also allows relatively unbiased sampling of other belowground C and nutrient stocks, including roots, coarse organic fragments, and rocks. We present a comprehensive methodology for sampling these pools with quantitative pits and assess their accuracy, precision, effort, and sampling intensity as compared to other methods. At 14 forested sites in New Hampshire, nonsoil belowground pools (which other methods may omit, double-count, or undercount) accounted for upward of 25% of total belowground C and N stocks: coarse material accounted for 4 and 1% of C and N in the O horizon; roots were 11 and 4% of C and N in the O horizon and 10 and 3% of C and N in the B horizon; and soil adhering to rocks represented 5% of total B-horizon C and N. The top 50 cm of the C horizon contained the equivalent of 17% of B-horizon carbon and N. Sampling procedures should be carefully designed to avoid treating these important pools inconsistently. Quantitative soil pits have fewer sources of systematic error than coring methods; the main disadvantage is that because they are time-consuming and create a larger zone of disturbance, fewer observations can be made than with cores.