The glyoxal-to-formaldehyde ratio (<em>R</em><sub>GF</sub>) has been proposed as a proxy to distinguish sources of volatile organic compounds (VOCs) in the atmosphere. However, the interpretation of its variability remains uncertain because of the diverse processes that affect VOC emissions and chemistry. In this study, we revisit the applicability and limitations of <em>R</em><sub>GF</sub> using multi-year ground-based MAX-DOAS measurements at four distinct sites: two biogenic (Orléans, France, and ATTO Tower, Brazil) and two anthropogenic (Athens, Greece, and Incheon, South Korea).</p> <p>The results show higher <em>R</em><sub>GF</sub> in anthropogenic environments and lower at biogenic sites. Seasonal <em>R</em><sub>GF</sub> patterns are broadly similar across sites, with reduced values in summer and enhanced values in winter, driven by HCHO variability. Diurnal cycles, caused by CHOCHO variability, are more pronounced at urban sites, with weekend effects of 10 %. Correlations between <em>R</em><sub>GF</sub> and NO<sub>2</sub> vary, even among anthropogenic stations, indicating the importance of local emission contributions.</p> <p>Additionally, our analysis shows that increasing temperatures leads to a decrease of <em>R</em><sub>GF</sub> by up to 1.9 percentage points across all sites, due to the more rapid increase of HCHO levels with temperature than CHOCHO.</p> <p>Moreover, we discuss four effects that reduce the comparability between different <em>R</em><sub>GF</sub> values: measurement volume, vertical sensitivity, time dependence, and the impact of averaging-ratioing order.</p> <p>Our findings suggest that ground-based remote sensing <em>R</em><sub>GF</sub> contains valuable diagnostic information about VOC source environments. However, its use as a universal proxy remains challenging, as our incomplete understanding of the various effects currently limits the reliable use of <em>R</em><sub>GF</sub> for VOC source attribution.