The detection of chemical substances excreted from the human body offers an attractive approach for non-invasive, early diagnostics of certain diseases. In this preliminary study, we proposed a susceptible optical sensor capable of quantitatively detecting ammonia from exhaled breath. The proposed sensor consists of nanoassembled ultrathin films composed of tetrakis(4-sulfophenyl)porphine (TSPP) and poly(diallyldimethylammonium chloride) (PDDA) deposited on quartz substrates using a layer-by-layer method. Measurement principles are based on the ammonia-induced absorbance changes at 489 (Soret band) and 702 nm (Q band), associated with the deprotonation of the J-aggregated TSPPs inside the film. Before exposure to breath, the PDDA/TSPP thin film was calibrated using known concentrations of ammonia gases with a projected detection limit of 102 ± 12 parts per billion (ppb). Calibrated sensor films were then exposed to human breath and urine samples to determine the ammonia concentration. Concentrations of exhaled ammonia are influenced significantly by the consumption of food or the amount of urea. Sensor response and maximum sensitivity, obtained from the absorbance changes induced by ammonia, were achieved by initial sensor exposure to HCl vapor. Previously reported procedures for the Helicobacter pylori (HELIC Ammonia Breath) test based on urea reaction with urease were reproduced using the proposed sensor. The observed behavior corresponded very well with the kinetics of the interactions between urea and urease, i.e., ammonia reached a maximum concentration approximately 5 min after the start of the reaction. A large-scale study involving 41 healthy volunteers in their 20s to 60s was successfully conducted to test the capabilities of the sensor to determine the concentration of exhaled ammonia. The concentration of ammonia for the healthy volunteers ranged between 0.3 and 1.5 ppm, with a mean value of ca. 520 ppb in the morning (before eating) and ca. 420 ppb in the afternoon (immediately after eating). These real-test mean values are meaningful when considered against the projected LOD.