The selection of proper tissues from formalin-fixed and paraffin-embedded tumors before diagnostic molecular testing is responsibility of the pathologist and represents a crucial step to produce reliable test results. The international guidelines suggest two cut-offs, one for the percentage and one for the number of tumor cells, in order to enrich the tumor content before DNA extraction. The aim of the present work was two-fold: to evaluate to what extent a low percentage or absolute number of tumor cells can be qualified for somatic mutation testing; and to determine how assay sensitivities can guide pathologists towards a better definition of morphology-based adequacy cut-offs. We tested 1797 tumor specimens from melanomas, colorectal and lung adenocarcinomas. Respectively, their BRAF, K-RAS and EGFR genes were analyzed at specific exons by mutation-enriched PCR, pyrosequencing, direct sequencing and real-time PCR methods. We demonstrate that poorly cellular specimens do not modify the frequency distribution of either mutated or wild-type DNA samples nor that of specific mutations. This observation suggests that currently recommended cut-offs for adequacy of specimens to be processed for molecular assays seem to be too much stringent in a laboratory context that performs highly sensitive routine analytical methods. In conclusion, new cut-offs are needed based on test sensitivities and documented tumor heterogeneity.