Quality assessment of the tissue stored in a tumor biobank is crucial because it is estimated that approximately 10% of the frozen samples are unsuitable for a molecular analysis mainly because of sampling problems in the tissue. We studied the value of imprint cytology (IC) versus frozen section to quantify necrosis and tumor cells in the tissue. The amount of tumor cells and necrosis was assessed by one pathologist on the frozen sections and ICs independently on 100 consecutive tumor samples. It was expressed as a percentage on frozen sections and on a four-level semiquantitative scale for IC (0 to 3+). Overall agreement between the quantity of tumor cells on IC and on frozen section was fair (Kappa = 0.23). Sensitivity and specificity of IC to detect the absence of tumor cell on the frozen section were 57% (4/7) and 98% (91/93), respectively. Overall agreement between necrosis quantification on IC and on frozen section was substantial (Kappa = 0.66).Sensitivity and specificity of IC to detect significant necrosis (defined as more than 30% necrosis) were 100% (3/3) and 98% (95/97), respectively. We show that IC is efficient to semiquantify necrosis in a tumor sample and to detect significant necrosis. IC seems to be less efficient to quantify tumor cells.