Conventional morphology-based identification is commonly used for routine assessment of freshwater ecosystems. However, cost and time efficient techniques such as high-throughput sequencing (HTS) based approaches may resolve the constraints encountered in conducting morphology-based surveys. Here, we characterized stream macroinvertebrate species diversity and community composition via metabarcoding and morphological analysis from environmental samples collected from the Shigenobu River Basin in Ehime Prefecture, Japan. We compared diversity metrics and assessed both approaches’ ability to evaluate the relationship between macroinvertebrate community and environmental variables. In total, we morphologically identified 45 taxa (3 families, six subfamilies, 31 genera, and five species) from 8,276 collected individuals from ten study sites. We detected 44 species by metabarcoding, with 35 species collapsed into 11 groups matching the morphologically identified taxa. A significant positive correlation between logged depth (number of HTS reads) and abundance of morphological taxa was observed, which implied that quantitative data could be used for subsequent analyses. We recovered a considerably high rate of relative abundance detection of the morphologically identified samples. Recovery of samples by incidence or presence/absence count were considerably low, with a high rate of false-negative detection specifically for species with scarce representation in the community sample. Given the low taxonomic resolution of the morphological assignment in this study, we report that metabarcoding does not reflect the majority of the species naturally occurring in our site, which could further be proven by performing refined morphological assessment of the samples. However, abundance-based detection proved to be efficient with 92% of the individuals correctly demonstrated. We conclude that DNA metabarcoding provides a practical and cost-effective approach especially for rapid biological monitoring of freshwater macroinvertebrate communities, but further improvements in the detection of scarce samples should be considered to increase the sensitivity of detecting most, if not all, of the species present in the environment.