Abstract. Since non-methane hydrocarbons (NMHCs) include numerous species of volatile organic compounds (VOCs), they are the only indicators that can characterize the total amount of VOCs in ambient air. More than 90 % of NMHC instruments in the market use the indirect method for NMHC determination, utilizing a gas chromatograph to determine the concentrations of total hydrocarbons (THCs) and methane in air. However, we found indirect NMHC measurements incorrectly characterized the low NMHC concentrations in the urban air. These measurements were hindered by the uncertain changes in the errors associated with low THC concentrations, humidity and macroscopic substances (oxygen) in the atmosphere. In this study, we conducted intercomparisons between 10 instruments in the market using different operation principles; among these, two used the direct method, and eight used the indirect method. For the indirect method, experiments showed that when the NMHC concentration was 50 ppb C, the errors in the measurement results were 32 % (CH4) and 98 % (THC) when compared to theoretical values. The oxygen content in the sample gas directly affected the response of the THC. The effect of increased oxygen content on the analysis of oxygen-containing substances was more significant than that of the increase in hydrocarbon substances. The response of dry zero was blank, and the response value increased non-linearly with relative humidity (RH). RH had a great impact on low NMHC concentrations of nearly 50 %. For the direct method, although various flow designs were applied for different instruments, the results indicated that they generally showed lower interference and higher accuracy for the ambient NMHC measurements. With the online direct methods, we obtained more reliable data and characteristics for low-concentration NMHCs in downtown Shanghai. The median of hourly data was 163.1 ppb C, and the highest ratio of the maximum and minimum concentrations of NMHCs reached 9.56 in a single day.