Understanding the oxidative reactivity of nanoparticles (NPs; <100 nm) could substantially contribute to explaining their toxicity. We attempted to refine the use of 2′7-dichlorodihydrofluorescein (DCFH) to characterize NP generation of reactive oxygen species (ROS). Several fluorescent probes have been applied to testing oxidative reactivity, but despite DCFH being one of the most popular for the detection of ROS, when it has been applied to NPs there have been an unexplainably wide variability in results. Without a uniform methodology, validating even robust results is impossible. This study, therefore, identified sources of conflicting results and investigated ways of reducing occurrence of artificial results. Existing techniques were tested and combined (using their most desirable features) to form a more reliable method for the measurement of NP reactivity in aqueous dispersions. We also investigated suitable sample ranges necessary to determine generation of ROS. Specifically, ultrafiltration and time-resolved scan absorbance spectra were used to study possible optical interference when using high sample concentrations. Robust results were achieved at a 5 µM DCFH working solution with 0.5 unit/mL horseradish peroxidase (HRP) dissolved in ethanol. Sonication in DCFH-HRP working solution provided more stable data with a relatively clean background. Optimal particle concentration depends on the type of NP and in general was in the µg/mL range. Major reasons for previously reported conflicting results due to interference were different experimental approaches and NP sample concentrations. The protocol presented here could form the basis of a standardized method for applying DCFH to detect generation of ROS by NPs.