The resting membrane potential and membrane potential changes are key to neuronal excitability and many other physiological processes such as cell-cycle progression and transmembrane transport. Thus, measurement of the membrane potential is essential for understanding cellular physiology. All-optical electrophysiology, using light to stimulate and record membrane potential changes, emerges as a new research field. Genetically-encoded fluorescent voltage indicators (GEVIs) are membrane proteins that change their fluorescence upon alteration of the electrical membrane potential. They are less invasive than patch clamp and supposedly less toxic than voltage-sensitive dyes. Currently available GEVIs, however, only provide a relative change in fluorescence as output, i. e., they do not report the absolute membrane potential. Here we introduce novel ratiometric GEVIs (rArc) based on the membrane-delimited green fluorescent ArcLightQ239, which changes its fluorescence intensity upon membrane potential variation (1). To calibrate the green voltage-sensitive signal of ArclightQ239, we fused variants of red fluorescent proteins (derivatives of mCherry and mKate2) to the intracellular N terminus of ArclightQ239. Alternatively, extracellular localization of the red fluorescent protein was achieved via an additional transmembrane helix. While the green signal diminished by about 44.0 ± 1.4% (mean ± S.E.M.) upon depolarisation from -120 to +60 mV (n = 20, for the mCherry variant rArc2C), the red fluorescence of all rArc variants remained constant in the same voltage range. The kinetics of the green fluorescence change of all rArc variants following a depolarization from -140 mV to 20 mV were similar to that of ArcLightQ239 (double-exponential fit: tau1 = 20 ± 2ms, tau2 = 173 ± 15 ms, Amp1/Amp2 = 0.83 ± 0.18, 23 °C, n = 6). 24 h after transfection in HEK293T cells, red and green fluorescence signals were colocalized. As observed for other red fluorescent proteins (2), intracellular red spots occurred after 48 h leading to a dysbalance of red and green signals. We therefore ranked rArc variants according to the homogeneity of red/green distribution in cells and the best fluorescence ratio reproducibility when assayed under voltage-clamp conditions. While the rArc variants, like ArcLightQ239 itself, are too slow for resolving fast action potentials, they promise great potential in characterizing the membrane potentials of resting cells. For example, rArc2C-based fluorescence ratio (Fgreen/Fred) in HEK293T cells significantly (U test, p < 0.001) increased from 1.42 ± 0.03 (n = 173) to 1.90 ± 0.06 (n = 114) when the inwardly rectifying and, hence, hyperpolarizing K+ channel mKir2.1 (3) was coexpressed. rArc variants will be examined for automated fluorescence recordings and for high-throughput applications such as fluorescence-activated cell sorting (FACS) and fluorescence imaging plate reader (FLIPR) (4) assays.