A large percentage of species richness in natural communities comprises rare taxa. These are often considered noisy and redundant to data analysis and are thus often excluded from ecological studies. We assessed the effect of these exclusions on biomonitoring of stream macroinvertebrates in Aotearoa New Zealand. We used macroinvertebrate data from 64 streams and rivers around the country, collected in 2005 as part of the National River Water Quality Network. We excluded rare taxa based on both site-specific and study-wide criteria, that set limits for inclusion of taxa based on the absolute and relative abundance, occurrence frequency, subsamples and species abundance ranking. The Macroinvertebrate Community Index was heavily affected, misclassifying numerous streams, upgrading the water quality classification for some and downgrading it for others. The biggest changes occurred using relative abundance criteria, which retained only a few, very abundant and tolerant taxa, downgrading the average index value for excellent, good and fair quality sites. The Quantitative MCI was not affected much, because of its numerical basis. In general, community structure correlated strongly between assemblages with different types of rare taxa excluded. The community structure of the quality class groups defined by each index were clearly distinguished after rare taxa exclusion, apart from some site-specific criteria. The relationship between MCI and nutrient concentrations (NOx and DRP) was weakened by extensive rare taxa exclusion. Study-wide relative abundance criteria excluded rare sensitive taxa, which could act as early warning indicators of increasing enrichment. The QMCI-nutrient relationship was not affected. The exclusion of rare taxa did not offer any significant advantages to biomonitoring and instead often led to misclassifications of the ecological quality of streams, with potentially severe implications for the distribution of management and restoration funds.