Introduction Formaldehyde is a carcinogenic pollutant omnipresent in indoor air due to emission from wood-based furniture, textiles, and paints [1]. But it is formed also outdoors from biofuel combustion, forest fires and photooxidation of smog [2]. Therefore, its exposure is strictly regulated worldwide (e.g., World Health Organization < 80 ppb for short-term exposure). However, reliable and portable devices for on-site monitoring do not exist. Currently, formaldehyde emission testing is carried out by sampling via sorption tubes and subsequent off-site laboratory analysis by gas chromatography. This is expensive, time-consuming, and thus not applicable for distributed indoor air quality monitoring or personal exposure assessment. Chemical sensors are promising for these applications as they are compact, low-cost and simple-in-use. Current sensors, however, typically fail to fulfill the necessary sensitivity and selectivity requirements to detect relevant formaldehyde concentrations below 100 parts per billion (ppb) in ambient air, which contains hundreds of interfering compounds. Here, we present a highly sensitive, selective and compact sensor array of flame-made metal-oxide sensors for real-time quantification of formaldehyde at 90% relative humidity (RH) in gas mixtures. Method Sensing nanoparticles of Pt-, Si-, Pd- and Ti-doped SnO2 are produced by flame-spray pyrolysis and directly deposited onto microsensor substrates by thermophoresis. The individual sensors are assembled to an array (E-nose) by wire-bonding them on a leadless chip carrier, which is placed inside a Teflon sensor chamber. Sensing films are heated to 400 °C with a Pt back heater and the E-nose is characterized in synthetic gas mixtures with a gas mixing setup at 90% RH [3]. The E-nose is tested in gas mixtures containing formaldehyde, acetone, ammonia and ethanol and a multilinear linear regression model is applied to obtain individual analyte concentrations [4]. Results and Conclusions The E-nose consists of four nanostructured and highly porous Pt-, Si-, Pd- and Ti-doped SnO2 sensing films precisely deposited onto silicon wafer-based microsubstrates by flame spray pyrolysis (Figure 1a, b). Flame-made sensing particles are deposited (black patterns) within the designated area of the interdigitated electrodes, while the heater and electrode pads are shielded by the shadow mask and not contaminated by particles. This technique is scalable and be readily applied to wafer-level processing (69 sensors in a single step [5]).The individual sensors offer a sensing film consisting of a highly porous and nanostructured network of nanoparticles (Figure 1b). As a result, they are highly sensitive for formaldehyde concentrations <10 ppb even at 90% RH with fast response dynamics (response time 140 s) as shown in Figure 2. Even concentrations as low as 5 ppb are detected with high signal-to-noise ratio >25. Each dopant induces different analyte selectivity, enabling selective detection of formaldehyde in gas mixtures by multivariate linear regression. We apply this for formaldehyde detection in complex gas mixtures of formaldehyde, acetone, NH3, and ethanol at realistic concentrations and humidity. Formaldehyde is thereby detected with an average error of ≤9 pbb at concentrations 30–180 ppb despite the much higher concentrated interferants.As a result, based on its compact size and low price, this device is promising for monitoring of formaldehyde in indoor and outdoor air. When interconnected, such next-generation low-cost detectors could enable distributed chemical recognition [6] for air quality in “smart” buildings and “future” cities.