The methods of unbiased (design-based) stereology are widely accepted as best practices for quantitative analysis of biological structures on stained tissue sections. Through systematic-random sampling (SRS) and unbiased geometric probes, modern stereology ensures accurate estimates of first- and second-order parameters that converge on the expected (true) values as sampling increases. Among the most common stereology parameter of bioscientific interest is total reference volume (Total V_ref in um³, mm³,…) using the Cavalieri method. The highly efficient approach is used to estimate Total V_ref as the product of 1) sum of areas on the cut surfaces of SRS sections (∑A_SRS , in um², mm²,…); and, 2) mean thickness (distance) between the surfaces (T_mean, in um³, mm³,…).  Accuracy of Total V_ref = ∑A_SRS •  T_mean depends on measurement of these quantities after equivalent section deformations, e.g., shrinkage, expansion, if any during processing. Here we show that the Total V_ref estimated using block advance for T_mean on ordinary processed sections introduces a systematic error (bias). The magnitude of this bias varies as a function of differential section shrinkage that occurs between cutting tissue sections at a uniform thickness (block advance) and the final thickness of fully stained and cover-slipped sections; that is, the product of ∑A_SRS and the block advance (T_mean) will calculate an imaginary volume since these values are quantified on fully processed and partially processed sections, respectively. As sampling continues, these imaginary volume estimates will not converge on any true tissue volume. Furthermore, the faulty use of block advance for T_mean, rather than the true final section thickness, also introduces error during estimation of total object number (Total N) using either the two-step optical disector method [Total N = N per unit volume (Nv) • Vref]; or, the equivalent optical fractionator method. A formula is provided that uses final section thickness forT_mean to correct volume estimates and their variation calculated with block advance for T_mean.