Oligonucleotide probes are increasingly being used to characterize natural microbial assemblages by Tyramide Signal Amplification-Fluorescent in situ Hybridization (TSA-FISH, or CAtalysed Reporter Deposition CARD-FISH). In view of the fast-growing rRNA databases, we re-evaluated the in silico specificity of eleven bacterial and eukaryotic probes and competitor frequently used for the quantification of marine picoplankton. We performed tests on cell cultures to decrease the risk for non-specific hybridization, before they are used on environmental samples. The probes were confronted to recent databases and hybridization conditions were tested against target strains matching perfectly with the probes, and against the closest non-target strains presenting one to four mismatches. We increased the hybridization stringency from 55 to 65% formamide for the Eub338+EubII+EubIII probe mix to be specific to the Eubacteria domain. In addition, we found that recent changes in the Gammaproteobacteria classification decreased the specificity of Gam42a probe, and that the Roseo536R and Ros537 probes were not specific to, and missed part of the Roseobacter clade. Changes in stringency conditions were important for bacterial probes; these induced, respectively, a significant increase, in Eubacteria and Roseobacter and no significant changes in Gammaproteobacteria concentrations from the investigated natural environment. We confirmed the eukaryotic probes original conditions, and propose the Euk1209+NChlo01+Chlo02 probe mix to target the largest picoeukaryotic diversity. Experiences acquired through these investigations leads us to propose the use of seven steps protocol for complete FISH probe specificity check-up to improve data quality in environmental studies.