This study aimed to compare the efficiency of two different sizes of the Surber sampler to assess benthic macroinvertebrates in headwater streams in two Amazonian regions. Two Surber samplers of different sizes were used, one measuring 20 × 20 cm and the other 30 × 30 cm, both with a 0.25-mm net. The number of replicates taken was 6 for the smaller sampler and 3 for the bigger one, maintaining approximately the same total sampled area. The study was carried out in 12 headwater streams with different environmental conditions. Biological metrics were calculated for each size at each site and compared within each stream health category. A two-way analysis of similarities test was performed to compare the community structure assessed by each method at each stream. A normalized sampling effort was used to quantify the number of samples required to correctly sample each site. The data did not show a significant difference between the two sizes regarding the taxonomic recruitment and the community structure sampled at each stream, but differences were found between the two sizes in dominance values and in Shannon index scores for the natural sites. Furthermore, the smaller Surber was able to assess 70% of the estimated richness in all sites, which suggests that it is better to assess benthic macroinvertebrates than the larger Surber. Moreover, the smaller Surber is easier to transport in the field, reducing the effort of the technician, and takes less time to sort the material collected with it, which can reduce the sample processing effort, therefore reducing the cost of the project.