Fluorescence in situ hybridization (FISH) using fluorochrome-labeled DNA oligonucleotide probes has been successfully applied for in situ detection of anaerobic ammonium oxidizing (anammox) bacteria. However, application of the standard FISH protocols to visualize anammox bacteria in biofilms from a laboratory-scale wastewater reactor produced only weak signals. Increased signal intensity was achieved either by modifying the standard FISH protocol, using peptide nucleic acid probes (PNA FISH), or applying horse radish peroxidase- (HRP-) labeled probes and subsequent catalyzed reporter deposition (CARD-FISH). A comparative analysis using anammox biofilm samples and suspended anammox biomass from different laboratory wastewater bioreactors revealed that the modified standard FISH protocol and the PNA FISH probes produced equally strong fluorescence signals on suspended biomass, but only weak signals were obtained with the biofilm samples. The probe signal intensities in the biofilm samples could be enhanced by enzymatic pre-treatment of fixed cells, and by increasing the hybridization time of the PNA FISH protocol. CARD-FISH always produced up to four-fold stronger fluorescent signals but unspecific fluorescence signals, likely caused by endogenous peroxidases as reported in several previous studies, compromised the results. Interference of the development of fluorescence intensity with endogenous peroxidases was also observed in cells of aerobic ammonium oxidizers like Nitrosomonas europea, and sulfate-reducers like Desulfobacter postgatei. Interestingly, no interference was observed with other peroxidase-positive microorganisms, suggesting that CARD-FISH is not only compromised by the mere presence of peroxidases. Pre-treatment of cells to inactivate peroxidase with HCl or autoclavation/pasteurization failed to inactive peroxidases, but H(2)O(2) significantly reduced endogenous peroxidase activity. However, for optimal inactivation, different H(2)O(2) concentrations and incubation time may be needed, depending on nature of sample and should therefore always be individually determined for each study.