DNA sequence-based studies have greatly advanced the microbiome field. However, contamination from reagents or laboratory kits can lead to biased results and impedes reproducibility. Knight and colleagues now show that cross-contamination between samples in 96-well plates occurs at high frequencies. Contamination occurred mainly during DNA extraction rather than during the PCR step, and plate-based methods had more well-to-well contamination compared with manual single tube extractions. Furthermore, well-to-well contamination was more prominent in samples with lower microbial biomass. Cross-contamination can lead to the introduction of additional bacteria to samples and thus affect diversity estimation. The findings also have implications for the approach of removing ‘background contaminants’ in negative controls during the analysis of sequencing results, as taxa that actually occur in the samples might be missed.