Continuous monitoring of ambient non-methane hydrocarbons (NMHCs) by automated gas chromatographs equipped with flame ionization detection (termed in-situ GC/FID) with hourly data resolution was instated in ozone non-attainment areas throughout Taiwan. Performance of these on-site in-situ GCs was validated by manual flask sampling, as well as by in-lab gas chromatography/mass spectrometry (GC/MS) analysis. More than 50 VOCs from C(2) to C(11) were analyzed by both methods. Ninety flask samples were collected in series near an in-situ GC monitoring station in order to closely compare with the in-situ measurements. Both time-series and scatter plots from the two methods are displayed and discussed. It was found that over-simplified, un-humidified single-point calibration leading to surface loss was responsible for the bias in the in-situ method, resulting in greater error in accuracy as VOC volatility decreased. Although this over-estimate of the concentrations was found across all target VOCs, both methods were able to consistently capture the variability of ambient VOCs, with R(2) values greater than 0.9 for most of the major VOCs. (C) 2009 Elsevier Ltd. All rights reserved.