Phytoplankton is a main primary producer, especially in pelagic ecosystems. Thus it needs to be well quantified for measuring its function and assessing water quality. To explore data quality of phytoplankton by the concentrated water sample method which has been widely applied in China and other developing countries, phytoplankton samples were collected from three reservoirs with distinct trophic levels and three treatment groups of experimental enclosures. The potential effects of the concentration ratios of water samples, counting volumes, replicates of water samples on the data quality were statistically analyzed. To balance the stability of data and counting efficiency under microscopy, we recommend to count 4 plates ×10 lattices for each plate. In the range of clearly and easily counting, increasing the concentration ratios of the water samples can improve precision and reliability of the quantitative data. When collecting multiple samples for counting is possible, especially those used for assessing the water quality, replicates of water samples are strongly recommended to be collected in order to reduce the standard deviation.